{
    "segments": [
        {
            "id": 1,
            "seek": 2898,
            "start": 3.72,
            "end": 14.66,
            "text": " Sorry about the no sound, people. I will figure this out. Every day,",
            "tokens": [
                50404,
                4919,
                466,
                264,
                572,
                1626,
                11,
                561,
                13,
                286,
                486,
                2573,
                341,
                484,
                13,
                2048,
                786,
                11,
                50860
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34966857370102045,
            "compression_ratio": 1.4488636363636365,
            "no_speech_prob": 0.186279296875,
            "words": [
                {
                    "start": 3.72,
                    "end": 4.02,
                    "word": " Sorry",
                    "probability": 0.449462890625
                },
                {
                    "start": 4.02,
                    "end": 4.22,
                    "word": " about",
                    "probability": 0.818359375
                },
                {
                    "start": 4.22,
                    "end": 4.38,
                    "word": " the",
                    "probability": 0.736328125
                },
                {
                    "start": 4.38,
                    "end": 4.52,
                    "word": " no",
                    "probability": 0.876953125
                },
                {
                    "start": 4.52,
                    "end": 4.76,
                    "word": " sound",
                    "probability": 0.80615234375
                },
                {
                    "start": 4.76,
                    "end": 4.94,
                    "word": ",",
                    "probability": 0.3271484375
                },
                {
                    "start": 4.94,
                    "end": 5.12,
                    "word": " people",
                    "probability": 0.91357421875
                },
                {
                    "start": 5.12,
                    "end": 5.68,
                    "word": ".",
                    "probability": 0.828125
                },
                {
                    "start": 5.68,
                    "end": 6.06,
                    "word": " I",
                    "probability": 0.94482421875
                },
                {
                    "start": 6.06,
                    "end": 6.44,
                    "word": " will",
                    "probability": 0.81591796875
                },
                {
                    "start": 6.44,
                    "end": 10.2,
                    "word": " figure",
                    "probability": 0.70263671875
                },
                {
                    "start": 10.2,
                    "end": 10.42,
                    "word": " this",
                    "probability": 0.880859375
                },
                {
                    "start": 10.42,
                    "end": 10.74,
                    "word": " out",
                    "probability": 0.8193359375
                },
                {
                    "start": 10.74,
                    "end": 11.34,
                    "word": ".",
                    "probability": 0.84619140625
                },
                {
                    "start": 11.96,
                    "end": 12.56,
                    "word": " Every",
                    "probability": 0.7333984375
                },
                {
                    "start": 12.56,
                    "end": 12.94,
                    "word": " day",
                    "probability": 0.89013671875
                },
                {
                    "start": 12.94,
                    "end": 14.66,
                    "word": ",",
                    "probability": 0.5068359375
                }
            ]
        },
        {
            "id": 2,
            "seek": 2898,
            "start": 14.66,
            "end": 25.96,
            "text": " it just decides to grab the wrong audio input. So let's go back to the beginning and see if",
            "tokens": [
                50956,
                309,
                445,
                14898,
                281,
                4444,
                264,
                2085,
                6278,
                4846,
                13,
                407,
                718,
                311,
                352,
                646,
                281,
                264,
                2863,
                293,
                536,
                498,
                51508
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34966857370102045,
            "compression_ratio": 1.4488636363636365,
            "no_speech_prob": 0.186279296875,
            "words": [
                {
                    "start": 14.66,
                    "end": 15.04,
                    "word": " it",
                    "probability": 0.8427734375
                },
                {
                    "start": 15.04,
                    "end": 15.32,
                    "word": " just",
                    "probability": 0.7119140625
                },
                {
                    "start": 15.32,
                    "end": 15.64,
                    "word": " decides",
                    "probability": 0.72607421875
                },
                {
                    "start": 15.64,
                    "end": 16.78,
                    "word": " to",
                    "probability": 0.80224609375
                },
                {
                    "start": 16.78,
                    "end": 17.48,
                    "word": " grab",
                    "probability": 0.71923828125
                },
                {
                    "start": 17.48,
                    "end": 17.72,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 17.72,
                    "end": 17.96,
                    "word": " wrong",
                    "probability": 0.85693359375
                },
                {
                    "start": 17.96,
                    "end": 18.32,
                    "word": " audio",
                    "probability": 0.90869140625
                },
                {
                    "start": 18.32,
                    "end": 18.68,
                    "word": " input",
                    "probability": 0.71875
                },
                {
                    "start": 18.68,
                    "end": 19.14,
                    "word": ".",
                    "probability": 0.90673828125
                },
                {
                    "start": 22.24,
                    "end": 22.84,
                    "word": " So",
                    "probability": 0.634765625
                },
                {
                    "start": 22.84,
                    "end": 23.08,
                    "word": " let's",
                    "probability": 0.8818359375
                },
                {
                    "start": 23.08,
                    "end": 23.22,
                    "word": " go",
                    "probability": 0.8955078125
                },
                {
                    "start": 23.22,
                    "end": 23.52,
                    "word": " back",
                    "probability": 0.8203125
                },
                {
                    "start": 23.52,
                    "end": 24.06,
                    "word": " to",
                    "probability": 0.82275390625
                },
                {
                    "start": 24.06,
                    "end": 24.92,
                    "word": " the",
                    "probability": 0.75732421875
                },
                {
                    "start": 24.92,
                    "end": 25.2,
                    "word": " beginning",
                    "probability": 0.810546875
                },
                {
                    "start": 25.2,
                    "end": 25.64,
                    "word": " and",
                    "probability": 0.66943359375
                },
                {
                    "start": 25.64,
                    "end": 25.82,
                    "word": " see",
                    "probability": 0.8154296875
                },
                {
                    "start": 25.82,
                    "end": 25.96,
                    "word": " if",
                    "probability": 0.81103515625
                }
            ]
        },
        {
            "id": 3,
            "seek": 2898,
            "start": 25.96,
            "end": 31.92,
            "text": " this works. Hi everyone. Today's talk is about machine learning privacy. The title of the talk",
            "tokens": [
                51508,
                341,
                1985,
                13,
                2421,
                1518,
                13,
                2692,
                311,
                751,
                307,
                466,
                3479,
                2539,
                11427,
                13,
                440,
                4876,
                295,
                264,
                751,
                51824
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34966857370102045,
            "compression_ratio": 1.4488636363636365,
            "no_speech_prob": 0.186279296875,
            "words": [
                {
                    "start": 25.96,
                    "end": 26.08,
                    "word": " this",
                    "probability": 0.525390625
                },
                {
                    "start": 26.08,
                    "end": 26.26,
                    "word": " works",
                    "probability": 0.712890625
                },
                {
                    "start": 26.26,
                    "end": 26.54,
                    "word": ".",
                    "probability": 0.875
                },
                {
                    "start": 26.54,
                    "end": 26.76,
                    "word": " Hi",
                    "probability": 0.5361328125
                },
                {
                    "start": 26.76,
                    "end": 27.1,
                    "word": " everyone",
                    "probability": 0.487060546875
                },
                {
                    "start": 27.1,
                    "end": 27.7,
                    "word": ".",
                    "probability": 0.65283203125
                },
                {
                    "start": 27.94,
                    "end": 28.54,
                    "word": " Today's",
                    "probability": 0.922119140625
                },
                {
                    "start": 28.54,
                    "end": 28.84,
                    "word": " talk",
                    "probability": 0.80517578125
                },
                {
                    "start": 28.84,
                    "end": 29.1,
                    "word": " is",
                    "probability": 0.90478515625
                },
                {
                    "start": 29.1,
                    "end": 29.5,
                    "word": " about",
                    "probability": 0.8173828125
                },
                {
                    "start": 29.5,
                    "end": 29.78,
                    "word": " machine",
                    "probability": 0.56689453125
                },
                {
                    "start": 29.78,
                    "end": 30.08,
                    "word": " learning",
                    "probability": 0.80615234375
                },
                {
                    "start": 30.08,
                    "end": 30.62,
                    "word": " privacy",
                    "probability": 0.82763671875
                },
                {
                    "start": 30.62,
                    "end": 30.82,
                    "word": ".",
                    "probability": 0.974609375
                },
                {
                    "start": 30.82,
                    "end": 30.92,
                    "word": " The",
                    "probability": 0.7978515625
                },
                {
                    "start": 30.92,
                    "end": 31.18,
                    "word": " title",
                    "probability": 0.93408203125
                },
                {
                    "start": 31.18,
                    "end": 31.42,
                    "word": " of",
                    "probability": 0.89404296875
                },
                {
                    "start": 31.42,
                    "end": 31.58,
                    "word": " the",
                    "probability": 0.79541015625
                },
                {
                    "start": 31.58,
                    "end": 31.92,
                    "word": " talk",
                    "probability": 0.78564453125
                }
            ]
        },
        {
            "id": 4,
            "seek": 5898,
            "start": 31.92,
            "end": 39.1,
            "text": " is secrets are lies, sharing is caring, privacy is theft. My name is Nahid Farhadi and I'm a",
            "tokens": [
                50376,
                307,
                14093,
                366,
                9134,
                11,
                5414,
                307,
                15365,
                11,
                11427,
                307,
                28508,
                13,
                1222,
                1315,
                307,
                13933,
                327,
                9067,
                71,
                5688,
                293,
                286,
                478,
                257,
                50704
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25555888649362785,
            "compression_ratio": 1.6217228464419475,
            "no_speech_prob": 0.00724029541015625,
            "words": [
                {
                    "start": 31.92,
                    "end": 32.24,
                    "word": " is",
                    "probability": 0.2374267578125
                },
                {
                    "start": 32.24,
                    "end": 32.92,
                    "word": " secrets",
                    "probability": 0.269775390625
                },
                {
                    "start": 32.92,
                    "end": 33.24,
                    "word": " are",
                    "probability": 0.28857421875
                },
                {
                    "start": 33.24,
                    "end": 33.64,
                    "word": " lies",
                    "probability": 0.72119140625
                },
                {
                    "start": 33.64,
                    "end": 33.94,
                    "word": ",",
                    "probability": 0.62353515625
                },
                {
                    "start": 33.94,
                    "end": 34.22,
                    "word": " sharing",
                    "probability": 0.89990234375
                },
                {
                    "start": 34.22,
                    "end": 34.5,
                    "word": " is",
                    "probability": 0.88916015625
                },
                {
                    "start": 34.5,
                    "end": 34.84,
                    "word": " caring",
                    "probability": 0.685546875
                },
                {
                    "start": 34.84,
                    "end": 35.12,
                    "word": ",",
                    "probability": 0.8759765625
                },
                {
                    "start": 35.12,
                    "end": 35.46,
                    "word": " privacy",
                    "probability": 0.83056640625
                },
                {
                    "start": 35.46,
                    "end": 35.8,
                    "word": " is",
                    "probability": 0.90185546875
                },
                {
                    "start": 35.8,
                    "end": 36.06,
                    "word": " theft",
                    "probability": 0.7958984375
                },
                {
                    "start": 36.06,
                    "end": 36.62,
                    "word": ".",
                    "probability": 0.943359375
                },
                {
                    "start": 36.92,
                    "end": 37.12,
                    "word": " My",
                    "probability": 0.90087890625
                },
                {
                    "start": 37.12,
                    "end": 37.34,
                    "word": " name",
                    "probability": 0.81640625
                },
                {
                    "start": 37.34,
                    "end": 37.48,
                    "word": " is",
                    "probability": 0.90283203125
                },
                {
                    "start": 37.48,
                    "end": 37.74,
                    "word": " Nahid",
                    "probability": 0.734619140625
                },
                {
                    "start": 37.74,
                    "end": 38.22,
                    "word": " Farhadi",
                    "probability": 0.8665364583333334
                },
                {
                    "start": 38.22,
                    "end": 38.5,
                    "word": " and",
                    "probability": 0.642578125
                },
                {
                    "start": 38.5,
                    "end": 38.74,
                    "word": " I'm",
                    "probability": 0.916259765625
                },
                {
                    "start": 38.84,
                    "end": 39.1,
                    "word": " a software",
                    "probability": 0.80908203125
                }
            ]
        },
        {
            "id": 5,
            "seek": 5898,
            "start": 39.1,
            "end": 45.52,
            "text": " software developer in Capital One. I will be giving this talk with my colleague Vincent Pham.",
            "tokens": [
                50704,
                4722,
                10754,
                294,
                21502,
                1485,
                13,
                286,
                486,
                312,
                2902,
                341,
                751,
                365,
                452,
                13532,
                28003,
                2623,
                335,
                13,
                50944
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25555888649362785,
            "compression_ratio": 1.6217228464419475,
            "no_speech_prob": 0.00724029541015625,
            "words": [
                {
                    "start": 39.1,
                    "end": 39.58,
                    "word": " developer",
                    "probability": 0.57958984375
                },
                {
                    "start": 39.58,
                    "end": 39.92,
                    "word": " in",
                    "probability": 0.8984375
                },
                {
                    "start": 39.92,
                    "end": 40.26,
                    "word": " Capital",
                    "probability": 0.8779296875
                },
                {
                    "start": 40.26,
                    "end": 40.54,
                    "word": " One",
                    "probability": 0.81640625
                },
                {
                    "start": 40.54,
                    "end": 40.7,
                    "word": ".",
                    "probability": 0.94091796875
                },
                {
                    "start": 40.7,
                    "end": 40.82,
                    "word": " I",
                    "probability": 0.9970703125
                },
                {
                    "start": 40.82,
                    "end": 41.0,
                    "word": " will",
                    "probability": 0.810546875
                },
                {
                    "start": 41.0,
                    "end": 41.16,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 41.16,
                    "end": 41.36,
                    "word": " giving",
                    "probability": 0.85498046875
                },
                {
                    "start": 41.36,
                    "end": 41.64,
                    "word": " this",
                    "probability": 0.8818359375
                },
                {
                    "start": 41.64,
                    "end": 41.92,
                    "word": " talk",
                    "probability": 0.8173828125
                },
                {
                    "start": 41.92,
                    "end": 42.14,
                    "word": " with",
                    "probability": 0.8134765625
                },
                {
                    "start": 42.14,
                    "end": 42.38,
                    "word": " my",
                    "probability": 0.91259765625
                },
                {
                    "start": 42.38,
                    "end": 42.72,
                    "word": " colleague",
                    "probability": 0.79052734375
                },
                {
                    "start": 42.72,
                    "end": 43.1,
                    "word": " Vincent",
                    "probability": 0.40576171875
                },
                {
                    "start": 43.1,
                    "end": 43.56,
                    "word": " Pham",
                    "probability": 0.8310546875
                },
                {
                    "start": 43.56,
                    "end": 44.12,
                    "word": ".",
                    "probability": 0.9560546875
                },
                {
                    "start": 45.32,
                    "end": 45.52,
                    "word": " The",
                    "probability": 0.78173828125
                }
            ]
        },
        {
            "id": 6,
            "seek": 5898,
            "start": 45.52,
            "end": 51.16,
            "text": " The outline of our talk is as below. We will give an introduction on the importance of privacy",
            "tokens": [
                51028,
                440,
                16387,
                295,
                527,
                751,
                307,
                382,
                2507,
                13,
                492,
                486,
                976,
                364,
                9339,
                322,
                264,
                7379,
                295,
                11427,
                51308
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25555888649362785,
            "compression_ratio": 1.6217228464419475,
            "no_speech_prob": 0.00724029541015625,
            "words": [
                {
                    "start": 45.52,
                    "end": 45.78,
                    "word": " outline",
                    "probability": 0.87548828125
                },
                {
                    "start": 45.78,
                    "end": 45.98,
                    "word": " of",
                    "probability": 0.9052734375
                },
                {
                    "start": 45.98,
                    "end": 46.18,
                    "word": " our",
                    "probability": 0.810546875
                },
                {
                    "start": 46.18,
                    "end": 46.42,
                    "word": " talk",
                    "probability": 0.78271484375
                },
                {
                    "start": 46.42,
                    "end": 46.6,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 46.6,
                    "end": 46.78,
                    "word": " as",
                    "probability": 0.87158203125
                },
                {
                    "start": 46.78,
                    "end": 47.0,
                    "word": " below",
                    "probability": 0.393798828125
                },
                {
                    "start": 47.0,
                    "end": 47.46,
                    "word": ".",
                    "probability": 0.7900390625
                },
                {
                    "start": 47.46,
                    "end": 47.78,
                    "word": " We",
                    "probability": 0.89501953125
                },
                {
                    "start": 47.78,
                    "end": 48.02,
                    "word": " will",
                    "probability": 0.81884765625
                },
                {
                    "start": 48.02,
                    "end": 48.22,
                    "word": " give",
                    "probability": 0.8134765625
                },
                {
                    "start": 48.22,
                    "end": 48.4,
                    "word": " an",
                    "probability": 0.8955078125
                },
                {
                    "start": 48.4,
                    "end": 48.84,
                    "word": " introduction",
                    "probability": 0.81787109375
                },
                {
                    "start": 48.84,
                    "end": 49.22,
                    "word": " on",
                    "probability": 0.88916015625
                },
                {
                    "start": 49.22,
                    "end": 49.32,
                    "word": " the",
                    "probability": 0.4990234375
                },
                {
                    "start": 49.32,
                    "end": 49.64,
                    "word": " importance",
                    "probability": 0.9140625
                },
                {
                    "start": 49.64,
                    "end": 49.9,
                    "word": " of",
                    "probability": 0.90283203125
                },
                {
                    "start": 49.9,
                    "end": 50.34,
                    "word": " privacy",
                    "probability": 0.82666015625
                },
                {
                    "start": 50.34,
                    "end": 51.16,
                    "word": " and",
                    "probability": 0.48583984375
                }
            ]
        },
        {
            "id": 7,
            "seek": 5898,
            "start": 51.16,
            "end": 55.26,
            "text": " and we will talk about privacy attacks, we propose some defense techniques,",
            "tokens": [
                51308,
                293,
                321,
                486,
                751,
                466,
                11427,
                8122,
                11,
                321,
                17421,
                512,
                7654,
                7512,
                11,
                51520
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25555888649362785,
            "compression_ratio": 1.6217228464419475,
            "no_speech_prob": 0.00724029541015625,
            "words": [
                {
                    "start": 51.16,
                    "end": 51.32,
                    "word": " we",
                    "probability": 0.89892578125
                },
                {
                    "start": 51.32,
                    "end": 51.48,
                    "word": " will",
                    "probability": 0.8046875
                },
                {
                    "start": 51.48,
                    "end": 51.72,
                    "word": " talk",
                    "probability": 0.8154296875
                },
                {
                    "start": 51.72,
                    "end": 52.02,
                    "word": " about",
                    "probability": 0.8359375
                },
                {
                    "start": 52.02,
                    "end": 52.42,
                    "word": " privacy",
                    "probability": 0.82568359375
                },
                {
                    "start": 52.42,
                    "end": 52.98,
                    "word": " attacks",
                    "probability": 0.76513671875
                },
                {
                    "start": 52.98,
                    "end": 53.16,
                    "word": ",",
                    "probability": 0.57080078125
                },
                {
                    "start": 53.16,
                    "end": 53.3,
                    "word": " we",
                    "probability": 0.8779296875
                },
                {
                    "start": 53.3,
                    "end": 53.7,
                    "word": " propose",
                    "probability": 0.7509765625
                },
                {
                    "start": 53.7,
                    "end": 54.08,
                    "word": " some",
                    "probability": 0.8125
                },
                {
                    "start": 54.08,
                    "end": 54.48,
                    "word": " defense",
                    "probability": 0.8095703125
                },
                {
                    "start": 54.48,
                    "end": 54.96,
                    "word": " techniques",
                    "probability": 0.8427734375
                },
                {
                    "start": 54.96,
                    "end": 55.12,
                    "word": ",",
                    "probability": 0.425537109375
                },
                {
                    "start": 55.12,
                    "end": 55.26,
                    "word": " and",
                    "probability": 0.89501953125
                }
            ]
        },
        {
            "id": 8,
            "seek": 5898,
            "start": 55.26,
            "end": 60.28,
            "text": " and finally we will show a demo of an attack as well as defense techniques.",
            "tokens": [
                51520,
                293,
                2721,
                321,
                486,
                855,
                257,
                10723,
                295,
                364,
                2690,
                382,
                731,
                382,
                7654,
                7512,
                13,
                51756
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25555888649362785,
            "compression_ratio": 1.6217228464419475,
            "no_speech_prob": 0.00724029541015625,
            "words": [
                {
                    "start": 55.26,
                    "end": 55.66,
                    "word": " finally",
                    "probability": 0.8046875
                },
                {
                    "start": 55.66,
                    "end": 55.86,
                    "word": " we",
                    "probability": 0.62353515625
                },
                {
                    "start": 55.86,
                    "end": 56.04,
                    "word": " will",
                    "probability": 0.81884765625
                },
                {
                    "start": 56.04,
                    "end": 56.28,
                    "word": " show",
                    "probability": 0.869140625
                },
                {
                    "start": 56.44,
                    "end": 56.7,
                    "word": " a demo",
                    "probability": 0.7470703125
                },
                {
                    "start": 56.7,
                    "end": 57.28,
                    "word": " of",
                    "probability": 0.892578125
                },
                {
                    "start": 57.28,
                    "end": 57.6,
                    "word": " an",
                    "probability": 0.9072265625
                },
                {
                    "start": 57.6,
                    "end": 58.06,
                    "word": " attack",
                    "probability": 0.82958984375
                },
                {
                    "start": 58.06,
                    "end": 58.32,
                    "word": " as",
                    "probability": 0.814453125
                },
                {
                    "start": 58.32,
                    "end": 58.52,
                    "word": " well",
                    "probability": 0.8779296875
                },
                {
                    "start": 58.52,
                    "end": 58.84,
                    "word": " as",
                    "probability": 0.9111328125
                },
                {
                    "start": 58.84,
                    "end": 59.24,
                    "word": " defense",
                    "probability": 0.8349609375
                },
                {
                    "start": 59.24,
                    "end": 59.76,
                    "word": " techniques",
                    "probability": 0.8466796875
                },
                {
                    "start": 59.76,
                    "end": 60.28,
                    "word": ".",
                    "probability": 0.99853515625
                }
            ]
        },
        {
            "id": 9,
            "seek": 8532,
            "start": 61.92,
            "end": 68.3,
            "text": " Machine learning needs data. So what if this data has sensitive information like medical data",
            "tokens": [
                50364,
                22155,
                2539,
                2203,
                1412,
                13,
                407,
                437,
                498,
                341,
                1412,
                575,
                9477,
                1589,
                411,
                4625,
                1412,
                50684
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23028273986918585,
            "compression_ratio": 1.6652173913043478,
            "no_speech_prob": 0.00044798851013183594,
            "words": [
                {
                    "start": 61.92,
                    "end": 62.32,
                    "word": " Machine",
                    "probability": 0.403076171875
                },
                {
                    "start": 62.32,
                    "end": 62.68,
                    "word": " learning",
                    "probability": 0.62353515625
                },
                {
                    "start": 62.68,
                    "end": 63.02,
                    "word": " needs",
                    "probability": 0.75732421875
                },
                {
                    "start": 63.02,
                    "end": 63.46,
                    "word": " data",
                    "probability": 0.81982421875
                },
                {
                    "start": 63.46,
                    "end": 64.04,
                    "word": ".",
                    "probability": 0.7412109375
                },
                {
                    "start": 64.04,
                    "end": 64.36,
                    "word": " So",
                    "probability": 0.8291015625
                },
                {
                    "start": 64.36,
                    "end": 64.66,
                    "word": " what",
                    "probability": 0.7265625
                },
                {
                    "start": 64.66,
                    "end": 64.82,
                    "word": " if",
                    "probability": 0.908203125
                },
                {
                    "start": 64.82,
                    "end": 65.04,
                    "word": " this",
                    "probability": 0.85595703125
                },
                {
                    "start": 65.04,
                    "end": 65.46,
                    "word": " data",
                    "probability": 0.81494140625
                },
                {
                    "start": 65.46,
                    "end": 65.88,
                    "word": " has",
                    "probability": 0.90478515625
                },
                {
                    "start": 65.88,
                    "end": 66.28,
                    "word": " sensitive",
                    "probability": 0.74755859375
                },
                {
                    "start": 66.28,
                    "end": 66.94,
                    "word": " information",
                    "probability": 0.71240234375
                },
                {
                    "start": 66.94,
                    "end": 67.5,
                    "word": " like",
                    "probability": 0.65625
                },
                {
                    "start": 67.5,
                    "end": 67.86,
                    "word": " medical",
                    "probability": 0.9169921875
                },
                {
                    "start": 67.86,
                    "end": 68.3,
                    "word": " data",
                    "probability": 0.78564453125
                }
            ]
        },
        {
            "id": 10,
            "seek": 8532,
            "start": 68.3,
            "end": 76.62,
            "text": " or financial data? Even if we trust the algorithms that are generating the model or model developers,",
            "tokens": [
                50684,
                420,
                4669,
                1412,
                30,
                2754,
                498,
                321,
                3361,
                264,
                14642,
                300,
                366,
                17746,
                264,
                2316,
                420,
                2316,
                8849,
                11,
                51100
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23028273986918585,
            "compression_ratio": 1.6652173913043478,
            "no_speech_prob": 0.00044798851013183594,
            "words": [
                {
                    "start": 68.3,
                    "end": 68.62,
                    "word": " or",
                    "probability": 0.88818359375
                },
                {
                    "start": 68.62,
                    "end": 69.12,
                    "word": " financial",
                    "probability": 0.798828125
                },
                {
                    "start": 69.12,
                    "end": 69.66,
                    "word": " data",
                    "probability": 0.78515625
                },
                {
                    "start": 69.66,
                    "end": 70.3,
                    "word": "?",
                    "probability": 0.58056640625
                },
                {
                    "start": 70.4,
                    "end": 70.66,
                    "word": " Even",
                    "probability": 0.7744140625
                },
                {
                    "start": 70.66,
                    "end": 70.96,
                    "word": " if",
                    "probability": 0.9013671875
                },
                {
                    "start": 70.96,
                    "end": 71.12,
                    "word": " we",
                    "probability": 0.89599609375
                },
                {
                    "start": 71.12,
                    "end": 71.4,
                    "word": " trust",
                    "probability": 0.8037109375
                },
                {
                    "start": 71.4,
                    "end": 71.58,
                    "word": " the",
                    "probability": 0.80029296875
                },
                {
                    "start": 71.58,
                    "end": 72.14,
                    "word": " algorithms",
                    "probability": 0.859375
                },
                {
                    "start": 72.14,
                    "end": 72.62,
                    "word": " that",
                    "probability": 0.8720703125
                },
                {
                    "start": 72.62,
                    "end": 73.1,
                    "word": " are",
                    "probability": 0.8984375
                },
                {
                    "start": 73.1,
                    "end": 74.34,
                    "word": " generating",
                    "probability": 0.7919921875
                },
                {
                    "start": 74.34,
                    "end": 74.6,
                    "word": " the",
                    "probability": 0.7470703125
                },
                {
                    "start": 74.6,
                    "end": 74.92,
                    "word": " model",
                    "probability": 0.8203125
                },
                {
                    "start": 74.92,
                    "end": 75.2,
                    "word": " or",
                    "probability": 0.845703125
                },
                {
                    "start": 75.2,
                    "end": 75.48,
                    "word": " model",
                    "probability": 0.79345703125
                },
                {
                    "start": 75.48,
                    "end": 76.12,
                    "word": " developers",
                    "probability": 0.69384765625
                },
                {
                    "start": 76.12,
                    "end": 76.62,
                    "word": ",",
                    "probability": 0.74560546875
                }
            ]
        },
        {
            "id": 11,
            "seek": 8532,
            "start": 76.62,
            "end": 82.08,
            "text": " sometimes just using the statistics by looking at the output of a model, we can find out some",
            "tokens": [
                51100,
                2171,
                445,
                1228,
                264,
                12523,
                538,
                1237,
                412,
                264,
                5598,
                295,
                257,
                2316,
                11,
                321,
                393,
                915,
                484,
                512,
                51372
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23028273986918585,
            "compression_ratio": 1.6652173913043478,
            "no_speech_prob": 0.00044798851013183594,
            "words": [
                {
                    "start": 76.62,
                    "end": 77.16,
                    "word": " sometimes",
                    "probability": 0.89453125
                },
                {
                    "start": 77.16,
                    "end": 77.52,
                    "word": " just",
                    "probability": 0.8212890625
                },
                {
                    "start": 77.52,
                    "end": 77.84,
                    "word": " using",
                    "probability": 0.80712890625
                },
                {
                    "start": 77.84,
                    "end": 78.04,
                    "word": " the",
                    "probability": 0.669921875
                },
                {
                    "start": 78.04,
                    "end": 78.56,
                    "word": " statistics",
                    "probability": 0.8642578125
                },
                {
                    "start": 78.56,
                    "end": 79.1,
                    "word": " by",
                    "probability": 0.72607421875
                },
                {
                    "start": 79.1,
                    "end": 79.36,
                    "word": " looking",
                    "probability": 0.80078125
                },
                {
                    "start": 79.36,
                    "end": 79.6,
                    "word": " at",
                    "probability": 0.91162109375
                },
                {
                    "start": 79.6,
                    "end": 79.76,
                    "word": " the",
                    "probability": 0.8134765625
                },
                {
                    "start": 79.76,
                    "end": 80.1,
                    "word": " output",
                    "probability": 0.81982421875
                },
                {
                    "start": 80.1,
                    "end": 80.3,
                    "word": " of",
                    "probability": 0.90869140625
                },
                {
                    "start": 80.52,
                    "end": 80.8,
                    "word": " a model",
                    "probability": 0.80224609375
                },
                {
                    "start": 80.8,
                    "end": 80.98,
                    "word": ",",
                    "probability": 0.5634765625
                },
                {
                    "start": 80.98,
                    "end": 81.1,
                    "word": " we",
                    "probability": 0.8984375
                },
                {
                    "start": 81.1,
                    "end": 81.32,
                    "word": " can",
                    "probability": 0.90185546875
                },
                {
                    "start": 81.32,
                    "end": 81.66,
                    "word": " find",
                    "probability": 0.82470703125
                },
                {
                    "start": 81.66,
                    "end": 81.88,
                    "word": " out",
                    "probability": 0.81396484375
                },
                {
                    "start": 81.88,
                    "end": 82.08,
                    "word": " some",
                    "probability": 0.80126953125
                }
            ]
        },
        {
            "id": 12,
            "seek": 8532,
            "start": 82.08,
            "end": 88.26,
            "text": " information about the input of the model. This type of threat to ML can be black box or white",
            "tokens": [
                51372,
                1589,
                466,
                264,
                4846,
                295,
                264,
                2316,
                13,
                639,
                2010,
                295,
                4734,
                281,
                21601,
                393,
                312,
                2211,
                2424,
                420,
                2418,
                51676
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23028273986918585,
            "compression_ratio": 1.6652173913043478,
            "no_speech_prob": 0.00044798851013183594,
            "words": [
                {
                    "start": 82.08,
                    "end": 82.66,
                    "word": " information",
                    "probability": 0.734375
                },
                {
                    "start": 82.66,
                    "end": 83.16,
                    "word": " about",
                    "probability": 0.81982421875
                },
                {
                    "start": 83.16,
                    "end": 83.34,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 83.34,
                    "end": 83.6,
                    "word": " input",
                    "probability": 0.87451171875
                },
                {
                    "start": 83.6,
                    "end": 83.78,
                    "word": " of",
                    "probability": 0.908203125
                },
                {
                    "start": 83.78,
                    "end": 83.92,
                    "word": " the",
                    "probability": 0.8173828125
                },
                {
                    "start": 83.92,
                    "end": 84.22,
                    "word": " model",
                    "probability": 0.78955078125
                },
                {
                    "start": 84.22,
                    "end": 84.86,
                    "word": ".",
                    "probability": 0.98388671875
                },
                {
                    "start": 84.88,
                    "end": 85.16,
                    "word": " This",
                    "probability": 0.70849609375
                },
                {
                    "start": 85.16,
                    "end": 85.42,
                    "word": " type",
                    "probability": 0.94189453125
                },
                {
                    "start": 85.42,
                    "end": 85.62,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 85.62,
                    "end": 85.88,
                    "word": " threat",
                    "probability": 0.43310546875
                },
                {
                    "start": 85.88,
                    "end": 86.16,
                    "word": " to",
                    "probability": 0.435546875
                },
                {
                    "start": 86.16,
                    "end": 86.54,
                    "word": " ML",
                    "probability": 0.796875
                },
                {
                    "start": 86.54,
                    "end": 86.86,
                    "word": " can",
                    "probability": 0.89453125
                },
                {
                    "start": 86.86,
                    "end": 87.08,
                    "word": " be",
                    "probability": 0.89599609375
                },
                {
                    "start": 87.08,
                    "end": 87.34,
                    "word": " black",
                    "probability": 0.72509765625
                },
                {
                    "start": 87.34,
                    "end": 87.68,
                    "word": " box",
                    "probability": 0.72412109375
                },
                {
                    "start": 87.68,
                    "end": 87.94,
                    "word": " or",
                    "probability": 0.9013671875
                },
                {
                    "start": 87.94,
                    "end": 88.26,
                    "word": " white",
                    "probability": 0.83935546875
                }
            ]
        },
        {
            "id": 13,
            "seek": 10950,
            "start": 88.26,
            "end": 94.26,
            "text": " box, meaning that the adversary doesn't necessarily need to have any access to the",
            "tokens": [
                50364,
                2424,
                11,
                3620,
                300,
                264,
                48222,
                1177,
                380,
                4725,
                643,
                281,
                362,
                604,
                2105,
                281,
                264,
                50660
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19125791516485094,
            "compression_ratio": 1.588235294117647,
            "no_speech_prob": 0.0005335807800292969,
            "words": [
                {
                    "start": 88.26,
                    "end": 88.6,
                    "word": " box",
                    "probability": 0.23486328125
                },
                {
                    "start": 88.6,
                    "end": 88.66,
                    "word": ",",
                    "probability": 0.654296875
                },
                {
                    "start": 88.66,
                    "end": 88.86,
                    "word": " meaning",
                    "probability": 0.81982421875
                },
                {
                    "start": 88.86,
                    "end": 89.22,
                    "word": " that",
                    "probability": 0.87939453125
                },
                {
                    "start": 89.22,
                    "end": 89.64,
                    "word": " the",
                    "probability": 0.75390625
                },
                {
                    "start": 89.64,
                    "end": 90.48,
                    "word": " adversary",
                    "probability": 0.822265625
                },
                {
                    "start": 90.48,
                    "end": 90.96,
                    "word": " doesn't",
                    "probability": 0.931396484375
                },
                {
                    "start": 90.96,
                    "end": 91.54,
                    "word": " necessarily",
                    "probability": 0.74755859375
                },
                {
                    "start": 91.54,
                    "end": 92.02,
                    "word": " need",
                    "probability": 0.8125
                },
                {
                    "start": 92.02,
                    "end": 92.2,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 92.2,
                    "end": 92.6,
                    "word": " have",
                    "probability": 0.88720703125
                },
                {
                    "start": 92.6,
                    "end": 93.04,
                    "word": " any",
                    "probability": 0.81884765625
                },
                {
                    "start": 93.04,
                    "end": 93.76,
                    "word": " access",
                    "probability": 0.78955078125
                },
                {
                    "start": 93.76,
                    "end": 94.1,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 94.1,
                    "end": 94.26,
                    "word": " the",
                    "probability": 0.8193359375
                }
            ]
        },
        {
            "id": 14,
            "seek": 10950,
            "start": 94.26,
            "end": 101.76,
            "text": " model specifics. So what is really privacy? Privacy means that if we are using some data",
            "tokens": [
                50660,
                2316,
                28454,
                13,
                407,
                437,
                307,
                534,
                11427,
                30,
                39691,
                2551,
                1355,
                300,
                498,
                321,
                366,
                1228,
                512,
                1412,
                51048
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19125791516485094,
            "compression_ratio": 1.588235294117647,
            "no_speech_prob": 0.0005335807800292969,
            "words": [
                {
                    "start": 94.26,
                    "end": 94.54,
                    "word": " model",
                    "probability": 0.81884765625
                },
                {
                    "start": 94.54,
                    "end": 95.14,
                    "word": " specifics",
                    "probability": 0.583984375
                },
                {
                    "start": 95.14,
                    "end": 95.78,
                    "word": ".",
                    "probability": 0.990234375
                },
                {
                    "start": 96.32,
                    "end": 96.64,
                    "word": " So",
                    "probability": 0.875
                },
                {
                    "start": 96.64,
                    "end": 96.9,
                    "word": " what",
                    "probability": 0.82666015625
                },
                {
                    "start": 96.9,
                    "end": 97.08,
                    "word": " is",
                    "probability": 0.89892578125
                },
                {
                    "start": 97.08,
                    "end": 97.34,
                    "word": " really",
                    "probability": 0.75439453125
                },
                {
                    "start": 97.34,
                    "end": 97.88,
                    "word": " privacy",
                    "probability": 0.81201171875
                },
                {
                    "start": 97.88,
                    "end": 98.14,
                    "word": "?",
                    "probability": 0.96533203125
                },
                {
                    "start": 98.14,
                    "end": 98.62,
                    "word": " Privacy",
                    "probability": 0.900634765625
                },
                {
                    "start": 98.62,
                    "end": 99.18,
                    "word": " means",
                    "probability": 0.89013671875
                },
                {
                    "start": 99.18,
                    "end": 99.54,
                    "word": " that",
                    "probability": 0.8818359375
                },
                {
                    "start": 99.54,
                    "end": 99.98,
                    "word": " if",
                    "probability": 0.8544921875
                },
                {
                    "start": 99.98,
                    "end": 100.16,
                    "word": " we",
                    "probability": 0.85400390625
                },
                {
                    "start": 100.16,
                    "end": 100.34,
                    "word": " are",
                    "probability": 0.9013671875
                },
                {
                    "start": 100.34,
                    "end": 100.88,
                    "word": " using",
                    "probability": 0.8291015625
                },
                {
                    "start": 100.88,
                    "end": 101.38,
                    "word": " some",
                    "probability": 0.82080078125
                },
                {
                    "start": 101.38,
                    "end": 101.76,
                    "word": " data",
                    "probability": 0.798828125
                }
            ]
        },
        {
            "id": 15,
            "seek": 10950,
            "start": 101.76,
            "end": 108.56,
            "text": " from our users for training a model, but just by looking at the output of the model,",
            "tokens": [
                51048,
                490,
                527,
                5022,
                337,
                3097,
                257,
                2316,
                11,
                457,
                445,
                538,
                1237,
                412,
                264,
                5598,
                295,
                264,
                2316,
                11,
                51376
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19125791516485094,
            "compression_ratio": 1.588235294117647,
            "no_speech_prob": 0.0005335807800292969,
            "words": [
                {
                    "start": 101.76,
                    "end": 102.22,
                    "word": " from",
                    "probability": 0.8212890625
                },
                {
                    "start": 102.22,
                    "end": 102.52,
                    "word": " our",
                    "probability": 0.81884765625
                },
                {
                    "start": 102.52,
                    "end": 103.04,
                    "word": " users",
                    "probability": 0.8486328125
                },
                {
                    "start": 103.04,
                    "end": 104.0,
                    "word": " for",
                    "probability": 0.88037109375
                },
                {
                    "start": 104.0,
                    "end": 104.54,
                    "word": " training",
                    "probability": 0.8271484375
                },
                {
                    "start": 104.8,
                    "end": 105.16,
                    "word": " a model",
                    "probability": 0.814453125
                },
                {
                    "start": 105.16,
                    "end": 105.76,
                    "word": ",",
                    "probability": 0.849609375
                },
                {
                    "start": 105.76,
                    "end": 105.94,
                    "word": " but",
                    "probability": 0.771484375
                },
                {
                    "start": 105.94,
                    "end": 106.2,
                    "word": " just",
                    "probability": 0.8896484375
                },
                {
                    "start": 106.2,
                    "end": 106.4,
                    "word": " by",
                    "probability": 0.9091796875
                },
                {
                    "start": 106.4,
                    "end": 106.7,
                    "word": " looking",
                    "probability": 0.80078125
                },
                {
                    "start": 106.7,
                    "end": 106.92,
                    "word": " at",
                    "probability": 0.912109375
                },
                {
                    "start": 106.92,
                    "end": 107.08,
                    "word": " the",
                    "probability": 0.818359375
                },
                {
                    "start": 107.08,
                    "end": 107.4,
                    "word": " output",
                    "probability": 0.80419921875
                },
                {
                    "start": 107.4,
                    "end": 107.6,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 107.6,
                    "end": 107.74,
                    "word": " the",
                    "probability": 0.818359375
                },
                {
                    "start": 107.74,
                    "end": 108.14,
                    "word": " model",
                    "probability": 0.82275390625
                },
                {
                    "start": 108.14,
                    "end": 108.56,
                    "word": ",",
                    "probability": 0.943359375
                }
            ]
        },
        {
            "id": 16,
            "seek": 10950,
            "start": 108.56,
            "end": 112.44,
            "text": " no one should be able to get information on the input of the model.",
            "tokens": [
                51376,
                572,
                472,
                820,
                312,
                1075,
                281,
                483,
                1589,
                322,
                264,
                4846,
                295,
                264,
                2316,
                13,
                51560
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19125791516485094,
            "compression_ratio": 1.588235294117647,
            "no_speech_prob": 0.0005335807800292969,
            "words": [
                {
                    "start": 108.56,
                    "end": 108.74,
                    "word": " no",
                    "probability": 0.90576171875
                },
                {
                    "start": 108.74,
                    "end": 108.96,
                    "word": " one",
                    "probability": 0.81640625
                },
                {
                    "start": 108.96,
                    "end": 109.18,
                    "word": " should",
                    "probability": 0.9287109375
                },
                {
                    "start": 109.18,
                    "end": 109.34,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 109.34,
                    "end": 109.62,
                    "word": " able",
                    "probability": 0.88232421875
                },
                {
                    "start": 109.62,
                    "end": 109.86,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 109.86,
                    "end": 110.08,
                    "word": " get",
                    "probability": 0.8994140625
                },
                {
                    "start": 110.08,
                    "end": 110.7,
                    "word": " information",
                    "probability": 0.73193359375
                },
                {
                    "start": 110.7,
                    "end": 111.1,
                    "word": " on",
                    "probability": 0.90380859375
                },
                {
                    "start": 111.1,
                    "end": 111.22,
                    "word": " the",
                    "probability": 0.7978515625
                },
                {
                    "start": 111.22,
                    "end": 111.48,
                    "word": " input",
                    "probability": 0.85546875
                },
                {
                    "start": 111.48,
                    "end": 111.66,
                    "word": " of",
                    "probability": 0.90380859375
                },
                {
                    "start": 111.66,
                    "end": 111.8,
                    "word": " the",
                    "probability": 0.80615234375
                },
                {
                    "start": 111.8,
                    "end": 112.12,
                    "word": " model",
                    "probability": 0.8134765625
                },
                {
                    "start": 112.12,
                    "end": 112.44,
                    "word": ".",
                    "probability": 0.99462890625
                }
            ]
        },
        {
            "id": 17,
            "seek": 13950,
            "start": 113.76,
            "end": 120.48,
            "text": " There is a trade-off between privacy and utility. If we have too much privacy,",
            "tokens": [
                50396,
                821,
                307,
                257,
                4923,
                12,
                4506,
                1296,
                11427,
                293,
                14877,
                13,
                759,
                321,
                362,
                886,
                709,
                11427,
                11,
                50704
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2625718390804598,
            "compression_ratio": 1.676056338028169,
            "no_speech_prob": 0.0003921985626220703,
            "words": [
                {
                    "start": 113.76,
                    "end": 114.44,
                    "word": " There",
                    "probability": 0.7001953125
                },
                {
                    "start": 114.44,
                    "end": 114.62,
                    "word": " is",
                    "probability": 0.9013671875
                },
                {
                    "start": 114.78,
                    "end": 115.04,
                    "word": " a trade",
                    "probability": 0.83984375
                },
                {
                    "start": 115.04,
                    "end": 115.24,
                    "word": "-off",
                    "probability": 0.648681640625
                },
                {
                    "start": 115.24,
                    "end": 115.66,
                    "word": " between",
                    "probability": 0.77587890625
                },
                {
                    "start": 115.66,
                    "end": 116.22,
                    "word": " privacy",
                    "probability": 0.79931640625
                },
                {
                    "start": 116.22,
                    "end": 116.78,
                    "word": " and",
                    "probability": 0.89599609375
                },
                {
                    "start": 116.78,
                    "end": 117.3,
                    "word": " utility",
                    "probability": 0.77685546875
                },
                {
                    "start": 117.3,
                    "end": 117.98,
                    "word": ".",
                    "probability": 0.93212890625
                },
                {
                    "start": 118.28,
                    "end": 118.66,
                    "word": " If",
                    "probability": 0.89892578125
                },
                {
                    "start": 118.66,
                    "end": 119.02,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 119.02,
                    "end": 119.3,
                    "word": " have",
                    "probability": 0.8759765625
                },
                {
                    "start": 119.3,
                    "end": 119.52,
                    "word": " too",
                    "probability": 0.8125
                },
                {
                    "start": 119.52,
                    "end": 119.76,
                    "word": " much",
                    "probability": 0.81884765625
                },
                {
                    "start": 119.76,
                    "end": 120.26,
                    "word": " privacy",
                    "probability": 0.8212890625
                },
                {
                    "start": 120.26,
                    "end": 120.48,
                    "word": ",",
                    "probability": 0.7119140625
                }
            ]
        },
        {
            "id": 18,
            "seek": 13950,
            "start": 120.48,
            "end": 125.92,
            "text": " then it will be very difficult to use the model. If we don't have much privacy, then",
            "tokens": [
                50704,
                550,
                309,
                486,
                312,
                588,
                2252,
                281,
                764,
                264,
                2316,
                13,
                759,
                321,
                500,
                380,
                362,
                709,
                11427,
                11,
                550,
                50980
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2625718390804598,
            "compression_ratio": 1.676056338028169,
            "no_speech_prob": 0.0003921985626220703,
            "words": [
                {
                    "start": 120.48,
                    "end": 120.62,
                    "word": " then",
                    "probability": 0.6650390625
                },
                {
                    "start": 120.62,
                    "end": 120.8,
                    "word": " it",
                    "probability": 0.89892578125
                },
                {
                    "start": 120.8,
                    "end": 121.0,
                    "word": " will",
                    "probability": 0.7763671875
                },
                {
                    "start": 121.0,
                    "end": 121.16,
                    "word": " be",
                    "probability": 0.90625
                },
                {
                    "start": 121.16,
                    "end": 121.48,
                    "word": " very",
                    "probability": 0.740234375
                },
                {
                    "start": 121.48,
                    "end": 121.88,
                    "word": " difficult",
                    "probability": 0.84814453125
                },
                {
                    "start": 121.88,
                    "end": 122.2,
                    "word": " to",
                    "probability": 0.904296875
                },
                {
                    "start": 122.2,
                    "end": 122.44,
                    "word": " use",
                    "probability": 0.81787109375
                },
                {
                    "start": 122.44,
                    "end": 122.64,
                    "word": " the",
                    "probability": 0.81103515625
                },
                {
                    "start": 122.64,
                    "end": 122.9,
                    "word": " model",
                    "probability": 0.8212890625
                },
                {
                    "start": 122.9,
                    "end": 123.16,
                    "word": ".",
                    "probability": 0.96142578125
                },
                {
                    "start": 123.16,
                    "end": 123.38,
                    "word": " If",
                    "probability": 0.90185546875
                },
                {
                    "start": 123.38,
                    "end": 123.62,
                    "word": " we",
                    "probability": 0.90576171875
                },
                {
                    "start": 123.62,
                    "end": 123.96,
                    "word": " don't",
                    "probability": 0.943603515625
                },
                {
                    "start": 123.96,
                    "end": 124.36,
                    "word": " have",
                    "probability": 0.88916015625
                },
                {
                    "start": 124.36,
                    "end": 124.66,
                    "word": " much",
                    "probability": 0.82177734375
                },
                {
                    "start": 124.66,
                    "end": 125.16,
                    "word": " privacy",
                    "probability": 0.81201171875
                },
                {
                    "start": 125.16,
                    "end": 125.58,
                    "word": ",",
                    "probability": 0.8935546875
                },
                {
                    "start": 125.58,
                    "end": 125.92,
                    "word": " then",
                    "probability": 0.7275390625
                }
            ]
        },
        {
            "id": 19,
            "seek": 13950,
            "start": 125.92,
            "end": 134.02,
            "text": " it will be useful, but then we will endanger our customers. PPML is there to define or find",
            "tokens": [
                51024,
                309,
                486,
                312,
                4420,
                11,
                457,
                550,
                321,
                486,
                31975,
                527,
                4581,
                13,
                430,
                18819,
                43,
                307,
                456,
                281,
                6964,
                420,
                915,
                51388
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2625718390804598,
            "compression_ratio": 1.676056338028169,
            "no_speech_prob": 0.0003921985626220703,
            "words": [
                {
                    "start": 125.92,
                    "end": 126.94,
                    "word": " it",
                    "probability": 0.87841796875
                },
                {
                    "start": 126.94,
                    "end": 127.08,
                    "word": " will",
                    "probability": 0.81787109375
                },
                {
                    "start": 127.08,
                    "end": 127.24,
                    "word": " be",
                    "probability": 0.90673828125
                },
                {
                    "start": 127.24,
                    "end": 127.7,
                    "word": " useful",
                    "probability": 0.79296875
                },
                {
                    "start": 127.7,
                    "end": 127.94,
                    "word": ",",
                    "probability": 0.6904296875
                },
                {
                    "start": 127.94,
                    "end": 128.16,
                    "word": " but",
                    "probability": 0.88525390625
                },
                {
                    "start": 128.16,
                    "end": 128.32,
                    "word": " then",
                    "probability": 0.69189453125
                },
                {
                    "start": 128.32,
                    "end": 128.46,
                    "word": " we",
                    "probability": 0.8837890625
                },
                {
                    "start": 128.46,
                    "end": 128.66,
                    "word": " will",
                    "probability": 0.826171875
                },
                {
                    "start": 128.66,
                    "end": 129.08,
                    "word": " endanger",
                    "probability": 0.8515625
                },
                {
                    "start": 129.08,
                    "end": 129.56,
                    "word": " our",
                    "probability": 0.826171875
                },
                {
                    "start": 129.56,
                    "end": 130.12,
                    "word": " customers",
                    "probability": 0.76123046875
                },
                {
                    "start": 130.12,
                    "end": 130.8,
                    "word": ".",
                    "probability": 0.990234375
                },
                {
                    "start": 130.88,
                    "end": 131.44,
                    "word": " PPML",
                    "probability": 0.6553548177083334
                },
                {
                    "start": 131.44,
                    "end": 131.7,
                    "word": " is",
                    "probability": 0.9013671875
                },
                {
                    "start": 131.7,
                    "end": 132.0,
                    "word": " there",
                    "probability": 0.85009765625
                },
                {
                    "start": 132.0,
                    "end": 132.34,
                    "word": " to",
                    "probability": 0.90478515625
                },
                {
                    "start": 132.34,
                    "end": 133.28,
                    "word": " define",
                    "probability": 0.9091796875
                },
                {
                    "start": 133.28,
                    "end": 133.54,
                    "word": " or",
                    "probability": 0.884765625
                },
                {
                    "start": 133.54,
                    "end": 134.02,
                    "word": " find",
                    "probability": 0.81689453125
                }
            ]
        },
        {
            "id": 20,
            "seek": 13950,
            "start": 134.02,
            "end": 142.36,
            "text": " the optimized point between utility and privacy. How can we sacrifice less by getting enough privacy?",
            "tokens": [
                51388,
                264,
                26941,
                935,
                1296,
                14877,
                293,
                11427,
                13,
                1012,
                393,
                321,
                11521,
                1570,
                538,
                1242,
                1547,
                11427,
                30,
                51804
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2625718390804598,
            "compression_ratio": 1.676056338028169,
            "no_speech_prob": 0.0003921985626220703,
            "words": [
                {
                    "start": 134.02,
                    "end": 134.24,
                    "word": " the",
                    "probability": 0.810546875
                },
                {
                    "start": 134.24,
                    "end": 134.64,
                    "word": " optimized",
                    "probability": 0.76220703125
                },
                {
                    "start": 134.64,
                    "end": 135.32,
                    "word": " point",
                    "probability": 0.8466796875
                },
                {
                    "start": 135.32,
                    "end": 135.78,
                    "word": " between",
                    "probability": 0.78369140625
                },
                {
                    "start": 135.78,
                    "end": 136.3,
                    "word": " utility",
                    "probability": 0.84619140625
                },
                {
                    "start": 136.3,
                    "end": 136.68,
                    "word": " and",
                    "probability": 0.89306640625
                },
                {
                    "start": 136.68,
                    "end": 137.16,
                    "word": " privacy",
                    "probability": 0.81591796875
                },
                {
                    "start": 137.16,
                    "end": 137.34,
                    "word": ".",
                    "probability": 0.92626953125
                },
                {
                    "start": 137.34,
                    "end": 137.6,
                    "word": " How",
                    "probability": 0.892578125
                },
                {
                    "start": 137.6,
                    "end": 138.02,
                    "word": " can",
                    "probability": 0.890625
                },
                {
                    "start": 138.02,
                    "end": 138.32,
                    "word": " we",
                    "probability": 0.88134765625
                },
                {
                    "start": 138.32,
                    "end": 138.88,
                    "word": " sacrifice",
                    "probability": 0.78271484375
                },
                {
                    "start": 138.88,
                    "end": 139.5,
                    "word": " less",
                    "probability": 0.873046875
                },
                {
                    "start": 139.5,
                    "end": 140.32,
                    "word": " by",
                    "probability": 0.890625
                },
                {
                    "start": 140.32,
                    "end": 140.88,
                    "word": " getting",
                    "probability": 0.89892578125
                },
                {
                    "start": 140.88,
                    "end": 141.52,
                    "word": " enough",
                    "probability": 0.783203125
                },
                {
                    "start": 141.52,
                    "end": 142.12,
                    "word": " privacy",
                    "probability": 0.822265625
                },
                {
                    "start": 142.12,
                    "end": 142.36,
                    "word": "?",
                    "probability": 0.9462890625
                }
            ]
        },
        {
            "id": 21,
            "seek": 16604,
            "start": 143.46,
            "end": 147.22,
            "text": " My colleague will be talking about privacy attacks next.",
            "tokens": [
                50404,
                1222,
                13532,
                486,
                312,
                1417,
                466,
                11427,
                8122,
                958,
                13,
                50552
            ],
            "temperature": 0.0,
            "avg_logprob": -0.270089290255592,
            "compression_ratio": 1.7408906882591093,
            "no_speech_prob": 0.0010156631469726562,
            "words": [
                {
                    "start": 143.46,
                    "end": 143.9,
                    "word": " My",
                    "probability": 0.316162109375
                },
                {
                    "start": 143.9,
                    "end": 144.34,
                    "word": " colleague",
                    "probability": 0.75732421875
                },
                {
                    "start": 144.34,
                    "end": 144.54,
                    "word": " will",
                    "probability": 0.806640625
                },
                {
                    "start": 144.54,
                    "end": 144.7,
                    "word": " be",
                    "probability": 0.908203125
                },
                {
                    "start": 144.7,
                    "end": 145.0,
                    "word": " talking",
                    "probability": 0.75439453125
                },
                {
                    "start": 145.0,
                    "end": 145.4,
                    "word": " about",
                    "probability": 0.83935546875
                },
                {
                    "start": 145.4,
                    "end": 145.86,
                    "word": " privacy",
                    "probability": 0.7607421875
                },
                {
                    "start": 145.86,
                    "end": 146.32,
                    "word": " attacks",
                    "probability": 0.7431640625
                },
                {
                    "start": 146.32,
                    "end": 146.78,
                    "word": " next",
                    "probability": 0.84228515625
                },
                {
                    "start": 146.78,
                    "end": 147.22,
                    "word": ".",
                    "probability": 0.98388671875
                }
            ]
        },
        {
            "id": 22,
            "seek": 16604,
            "start": 148.06,
            "end": 152.92,
            "text": " In terms of privacy attack, there are two main categories. One is the classical, where the",
            "tokens": [
                50628,
                682,
                2115,
                295,
                11427,
                2690,
                11,
                456,
                366,
                732,
                2135,
                10479,
                13,
                1485,
                307,
                264,
                13735,
                11,
                689,
                264,
                50856
            ],
            "temperature": 0.0,
            "avg_logprob": -0.270089290255592,
            "compression_ratio": 1.7408906882591093,
            "no_speech_prob": 0.0010156631469726562,
            "words": [
                {
                    "start": 148.06,
                    "end": 148.5,
                    "word": " In",
                    "probability": 0.84375
                },
                {
                    "start": 148.5,
                    "end": 148.74,
                    "word": " terms",
                    "probability": 0.7509765625
                },
                {
                    "start": 148.74,
                    "end": 148.88,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 148.88,
                    "end": 149.2,
                    "word": " privacy",
                    "probability": 0.8046875
                },
                {
                    "start": 149.2,
                    "end": 149.58,
                    "word": " attack",
                    "probability": 0.5966796875
                },
                {
                    "start": 149.58,
                    "end": 149.82,
                    "word": ",",
                    "probability": 0.88720703125
                },
                {
                    "start": 149.82,
                    "end": 149.92,
                    "word": " there",
                    "probability": 0.85302734375
                },
                {
                    "start": 149.92,
                    "end": 150.1,
                    "word": " are",
                    "probability": 0.89306640625
                },
                {
                    "start": 150.1,
                    "end": 150.34,
                    "word": " two",
                    "probability": 0.80224609375
                },
                {
                    "start": 150.34,
                    "end": 150.56,
                    "word": " main",
                    "probability": 0.8388671875
                },
                {
                    "start": 150.56,
                    "end": 151.08,
                    "word": " categories",
                    "probability": 0.7939453125
                },
                {
                    "start": 151.08,
                    "end": 151.4,
                    "word": ".",
                    "probability": 0.85791015625
                },
                {
                    "start": 151.4,
                    "end": 151.56,
                    "word": " One",
                    "probability": 0.88427734375
                },
                {
                    "start": 151.56,
                    "end": 151.76,
                    "word": " is",
                    "probability": 0.8896484375
                },
                {
                    "start": 151.76,
                    "end": 151.92,
                    "word": " the",
                    "probability": 0.8046875
                },
                {
                    "start": 151.92,
                    "end": 152.24,
                    "word": " classical",
                    "probability": 0.908203125
                },
                {
                    "start": 152.24,
                    "end": 152.5,
                    "word": ",",
                    "probability": 0.369384765625
                },
                {
                    "start": 152.5,
                    "end": 152.68,
                    "word": " where",
                    "probability": 0.89208984375
                },
                {
                    "start": 152.68,
                    "end": 152.92,
                    "word": " the",
                    "probability": 0.80615234375
                }
            ]
        },
        {
            "id": 23,
            "seek": 16604,
            "start": 152.92,
            "end": 157.28,
            "text": " private data is in the train, where it's in the raw set. And there are the ML-enabled one, which",
            "tokens": [
                50856,
                4551,
                1412,
                307,
                294,
                264,
                3847,
                11,
                689,
                309,
                311,
                294,
                264,
                8936,
                992,
                13,
                400,
                456,
                366,
                264,
                21601,
                12,
                268,
                8909,
                472,
                11,
                597,
                51080
            ],
            "temperature": 0.0,
            "avg_logprob": -0.270089290255592,
            "compression_ratio": 1.7408906882591093,
            "no_speech_prob": 0.0010156631469726562,
            "words": [
                {
                    "start": 152.92,
                    "end": 153.2,
                    "word": " private",
                    "probability": 0.8095703125
                },
                {
                    "start": 153.2,
                    "end": 153.4,
                    "word": " data",
                    "probability": 0.77734375
                },
                {
                    "start": 153.4,
                    "end": 153.56,
                    "word": " is",
                    "probability": 0.88916015625
                },
                {
                    "start": 153.56,
                    "end": 153.66,
                    "word": " in",
                    "probability": 0.85791015625
                },
                {
                    "start": 153.66,
                    "end": 153.76,
                    "word": " the",
                    "probability": 0.6640625
                },
                {
                    "start": 153.76,
                    "end": 153.98,
                    "word": " train",
                    "probability": 0.492919921875
                },
                {
                    "start": 153.98,
                    "end": 154.08,
                    "word": ",",
                    "probability": 0.69189453125
                },
                {
                    "start": 154.08,
                    "end": 154.26,
                    "word": " where",
                    "probability": 0.89990234375
                },
                {
                    "start": 154.26,
                    "end": 154.4,
                    "word": " it's",
                    "probability": 0.932861328125
                },
                {
                    "start": 154.4,
                    "end": 154.52,
                    "word": " in",
                    "probability": 0.9033203125
                },
                {
                    "start": 154.52,
                    "end": 154.64,
                    "word": " the",
                    "probability": 0.79833984375
                },
                {
                    "start": 154.64,
                    "end": 154.84,
                    "word": " raw",
                    "probability": 0.78955078125
                },
                {
                    "start": 154.84,
                    "end": 155.1,
                    "word": " set",
                    "probability": 0.88916015625
                },
                {
                    "start": 155.1,
                    "end": 155.44,
                    "word": ".",
                    "probability": 0.466796875
                },
                {
                    "start": 155.44,
                    "end": 155.58,
                    "word": " And",
                    "probability": 0.78466796875
                },
                {
                    "start": 155.58,
                    "end": 155.72,
                    "word": " there",
                    "probability": 0.74951171875
                },
                {
                    "start": 155.72,
                    "end": 155.9,
                    "word": " are",
                    "probability": 0.86865234375
                },
                {
                    "start": 155.9,
                    "end": 156.08,
                    "word": " the",
                    "probability": 0.80224609375
                },
                {
                    "start": 156.08,
                    "end": 156.3,
                    "word": " ML",
                    "probability": 0.82861328125
                },
                {
                    "start": 156.3,
                    "end": 156.6,
                    "word": "-enabled",
                    "probability": 0.7300618489583334
                },
                {
                    "start": 156.6,
                    "end": 156.94,
                    "word": " one",
                    "probability": 0.556640625
                },
                {
                    "start": 156.94,
                    "end": 157.04,
                    "word": ",",
                    "probability": 0.90087890625
                },
                {
                    "start": 157.04,
                    "end": 157.28,
                    "word": " which",
                    "probability": 0.912109375
                }
            ]
        },
        {
            "id": 24,
            "seek": 16604,
            "start": 157.28,
            "end": 162.02,
            "text": " I'll be focusing more on. In the next slide, you can see that there are several attack surfaces",
            "tokens": [
                51080,
                286,
                603,
                312,
                8416,
                544,
                322,
                13,
                682,
                264,
                958,
                4137,
                11,
                291,
                393,
                536,
                300,
                456,
                366,
                2940,
                2690,
                16130,
                51320
            ],
            "temperature": 0.0,
            "avg_logprob": -0.270089290255592,
            "compression_ratio": 1.7408906882591093,
            "no_speech_prob": 0.0010156631469726562,
            "words": [
                {
                    "start": 157.28,
                    "end": 157.56,
                    "word": " I'll",
                    "probability": 0.836669921875
                },
                {
                    "start": 157.56,
                    "end": 157.72,
                    "word": " be",
                    "probability": 0.90380859375
                },
                {
                    "start": 157.72,
                    "end": 158.04,
                    "word": " focusing",
                    "probability": 0.759765625
                },
                {
                    "start": 158.04,
                    "end": 158.4,
                    "word": " more",
                    "probability": 0.88232421875
                },
                {
                    "start": 158.4,
                    "end": 158.6,
                    "word": " on",
                    "probability": 0.91064453125
                },
                {
                    "start": 158.6,
                    "end": 159.04,
                    "word": ".",
                    "probability": 0.98974609375
                },
                {
                    "start": 159.26,
                    "end": 159.42,
                    "word": " In",
                    "probability": 0.86962890625
                },
                {
                    "start": 159.42,
                    "end": 159.52,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 159.52,
                    "end": 159.72,
                    "word": " next",
                    "probability": 0.857421875
                },
                {
                    "start": 159.72,
                    "end": 160.08,
                    "word": " slide",
                    "probability": 0.89794921875
                },
                {
                    "start": 160.08,
                    "end": 160.26,
                    "word": ",",
                    "probability": 0.9658203125
                },
                {
                    "start": 160.26,
                    "end": 160.42,
                    "word": " you",
                    "probability": 0.76318359375
                },
                {
                    "start": 160.42,
                    "end": 160.52,
                    "word": " can",
                    "probability": 0.85888671875
                },
                {
                    "start": 160.52,
                    "end": 160.64,
                    "word": " see",
                    "probability": 0.82080078125
                },
                {
                    "start": 160.64,
                    "end": 160.8,
                    "word": " that",
                    "probability": 0.8779296875
                },
                {
                    "start": 160.8,
                    "end": 160.98,
                    "word": " there",
                    "probability": 0.861328125
                },
                {
                    "start": 160.98,
                    "end": 161.1,
                    "word": " are",
                    "probability": 0.8720703125
                },
                {
                    "start": 161.1,
                    "end": 161.28,
                    "word": " several",
                    "probability": 0.6689453125
                },
                {
                    "start": 161.28,
                    "end": 161.58,
                    "word": " attack",
                    "probability": 0.716796875
                },
                {
                    "start": 161.58,
                    "end": 162.02,
                    "word": " surfaces",
                    "probability": 0.74462890625
                }
            ]
        },
        {
            "id": 25,
            "seek": 16604,
            "start": 162.02,
            "end": 168.98,
            "text": " that attackers can utilize to perform a privacy attack. The first is the physical domain,",
            "tokens": [
                51320,
                300,
                45129,
                393,
                16117,
                281,
                2042,
                257,
                11427,
                2690,
                13,
                440,
                700,
                307,
                264,
                4001,
                9274,
                11,
                51656
            ],
            "temperature": 0.0,
            "avg_logprob": -0.270089290255592,
            "compression_ratio": 1.7408906882591093,
            "no_speech_prob": 0.0010156631469726562,
            "words": [
                {
                    "start": 162.02,
                    "end": 162.54,
                    "word": " that",
                    "probability": 0.884765625
                },
                {
                    "start": 162.54,
                    "end": 163.18,
                    "word": " attackers",
                    "probability": 0.81298828125
                },
                {
                    "start": 163.18,
                    "end": 163.42,
                    "word": " can",
                    "probability": 0.890625
                },
                {
                    "start": 163.42,
                    "end": 163.8,
                    "word": " utilize",
                    "probability": 0.7724609375
                },
                {
                    "start": 163.8,
                    "end": 164.38,
                    "word": " to",
                    "probability": 0.9052734375
                },
                {
                    "start": 164.38,
                    "end": 165.42,
                    "word": " perform",
                    "probability": 0.75390625
                },
                {
                    "start": 165.6,
                    "end": 165.94,
                    "word": " a privacy",
                    "probability": 0.58935546875
                },
                {
                    "start": 165.94,
                    "end": 166.92,
                    "word": " attack",
                    "probability": 0.80419921875
                },
                {
                    "start": 166.92,
                    "end": 167.3,
                    "word": ".",
                    "probability": 0.998046875
                },
                {
                    "start": 167.3,
                    "end": 167.74,
                    "word": " The",
                    "probability": 0.81396484375
                },
                {
                    "start": 167.74,
                    "end": 167.92,
                    "word": " first",
                    "probability": 0.814453125
                },
                {
                    "start": 167.92,
                    "end": 168.04,
                    "word": " is",
                    "probability": 0.90869140625
                },
                {
                    "start": 168.04,
                    "end": 168.16,
                    "word": " the",
                    "probability": 0.7626953125
                },
                {
                    "start": 168.16,
                    "end": 168.38,
                    "word": " physical",
                    "probability": 0.8330078125
                },
                {
                    "start": 168.38,
                    "end": 168.8,
                    "word": " domain",
                    "probability": 0.88623046875
                },
                {
                    "start": 168.8,
                    "end": 168.98,
                    "word": ",",
                    "probability": 0.58740234375
                }
            ]
        },
        {
            "id": 26,
            "seek": 19242,
            "start": 168.98,
            "end": 174.4,
            "text": " in terms of network protection, it could be attack network traffic. Then there's the digital",
            "tokens": [
                50384,
                294,
                2115,
                295,
                3209,
                6334,
                11,
                309,
                727,
                312,
                2690,
                3209,
                6419,
                13,
                1396,
                456,
                311,
                264,
                4562,
                50632
            ],
            "temperature": 0.0,
            "avg_logprob": -0.24885320663452148,
            "compression_ratio": 1.7709923664122138,
            "no_speech_prob": 0.0014324188232421875,
            "words": [
                {
                    "start": 168.98,
                    "end": 169.32,
                    "word": " in",
                    "probability": 0.069580078125
                },
                {
                    "start": 169.32,
                    "end": 169.82,
                    "word": " terms",
                    "probability": 0.759765625
                },
                {
                    "start": 169.82,
                    "end": 170.02,
                    "word": " of",
                    "probability": 0.90576171875
                },
                {
                    "start": 170.02,
                    "end": 170.28,
                    "word": " network",
                    "probability": 0.81982421875
                },
                {
                    "start": 170.28,
                    "end": 170.8,
                    "word": " protection",
                    "probability": 0.0175933837890625
                },
                {
                    "start": 170.8,
                    "end": 171.46,
                    "word": ",",
                    "probability": 0.5390625
                },
                {
                    "start": 171.46,
                    "end": 171.7,
                    "word": " it",
                    "probability": 0.6044921875
                },
                {
                    "start": 171.7,
                    "end": 171.8,
                    "word": " could",
                    "probability": 0.81884765625
                },
                {
                    "start": 171.8,
                    "end": 171.9,
                    "word": " be",
                    "probability": 0.908203125
                },
                {
                    "start": 171.9,
                    "end": 172.2,
                    "word": " attack",
                    "probability": 0.279052734375
                },
                {
                    "start": 172.2,
                    "end": 172.54,
                    "word": " network",
                    "probability": 0.84765625
                },
                {
                    "start": 172.54,
                    "end": 173.04,
                    "word": " traffic",
                    "probability": 0.80126953125
                },
                {
                    "start": 173.04,
                    "end": 173.34,
                    "word": ".",
                    "probability": 0.6943359375
                },
                {
                    "start": 173.4,
                    "end": 173.88,
                    "word": " Then",
                    "probability": 0.81884765625
                },
                {
                    "start": 173.88,
                    "end": 174.12,
                    "word": " there's",
                    "probability": 0.89501953125
                },
                {
                    "start": 174.12,
                    "end": 174.2,
                    "word": " the",
                    "probability": 0.5400390625
                },
                {
                    "start": 174.2,
                    "end": 174.4,
                    "word": " digital",
                    "probability": 0.70947265625
                }
            ]
        },
        {
            "id": 27,
            "seek": 19242,
            "start": 174.4,
            "end": 180.14,
            "text": " representation, which is the TCP dump. There's also the machine learning model, which is just",
            "tokens": [
                50632,
                10290,
                11,
                597,
                307,
                264,
                48965,
                11430,
                13,
                821,
                311,
                611,
                264,
                3479,
                2539,
                2316,
                11,
                597,
                307,
                445,
                50928
            ],
            "temperature": 0.0,
            "avg_logprob": -0.24885320663452148,
            "compression_ratio": 1.7709923664122138,
            "no_speech_prob": 0.0014324188232421875,
            "words": [
                {
                    "start": 174.4,
                    "end": 174.92,
                    "word": " representation",
                    "probability": 0.7890625
                },
                {
                    "start": 174.92,
                    "end": 175.46,
                    "word": ",",
                    "probability": 0.8701171875
                },
                {
                    "start": 175.46,
                    "end": 175.58,
                    "word": " which",
                    "probability": 0.91259765625
                },
                {
                    "start": 175.58,
                    "end": 175.7,
                    "word": " is",
                    "probability": 0.89892578125
                },
                {
                    "start": 175.7,
                    "end": 175.84,
                    "word": " the",
                    "probability": 0.796875
                },
                {
                    "start": 175.84,
                    "end": 176.2,
                    "word": " TCP",
                    "probability": 0.826171875
                },
                {
                    "start": 176.2,
                    "end": 176.6,
                    "word": " dump",
                    "probability": 0.80712890625
                },
                {
                    "start": 176.6,
                    "end": 177.08,
                    "word": ".",
                    "probability": 0.943359375
                },
                {
                    "start": 177.12,
                    "end": 177.48,
                    "word": " There's",
                    "probability": 0.89599609375
                },
                {
                    "start": 177.48,
                    "end": 177.96,
                    "word": " also",
                    "probability": 0.70458984375
                },
                {
                    "start": 177.96,
                    "end": 178.18,
                    "word": " the",
                    "probability": 0.8115234375
                },
                {
                    "start": 178.18,
                    "end": 178.44,
                    "word": " machine",
                    "probability": 0.70361328125
                },
                {
                    "start": 178.44,
                    "end": 178.68,
                    "word": " learning",
                    "probability": 0.81494140625
                },
                {
                    "start": 178.68,
                    "end": 179.08,
                    "word": " model",
                    "probability": 0.80859375
                },
                {
                    "start": 179.08,
                    "end": 179.36,
                    "word": ",",
                    "probability": 0.9208984375
                },
                {
                    "start": 179.36,
                    "end": 179.6,
                    "word": " which",
                    "probability": 0.9111328125
                },
                {
                    "start": 179.6,
                    "end": 179.84,
                    "word": " is",
                    "probability": 0.9013671875
                },
                {
                    "start": 179.84,
                    "end": 180.14,
                    "word": " just",
                    "probability": 0.88232421875
                }
            ]
        },
        {
            "id": 28,
            "seek": 19242,
            "start": 180.14,
            "end": 186.84,
            "text": " the model itself, where you have an input and do a prediction on the output. In terms of this,",
            "tokens": [
                50928,
                264,
                2316,
                2564,
                11,
                689,
                291,
                362,
                364,
                4846,
                293,
                360,
                257,
                17630,
                322,
                264,
                5598,
                13,
                682,
                2115,
                295,
                341,
                11,
                51252
            ],
            "temperature": 0.0,
            "avg_logprob": -0.24885320663452148,
            "compression_ratio": 1.7709923664122138,
            "no_speech_prob": 0.0014324188232421875,
            "words": [
                {
                    "start": 180.14,
                    "end": 180.58,
                    "word": " the",
                    "probability": 0.81201171875
                },
                {
                    "start": 180.58,
                    "end": 180.8,
                    "word": " model",
                    "probability": 0.80322265625
                },
                {
                    "start": 180.8,
                    "end": 181.24,
                    "word": " itself",
                    "probability": 0.77734375
                },
                {
                    "start": 181.24,
                    "end": 181.3,
                    "word": ",",
                    "probability": 0.51416015625
                },
                {
                    "start": 181.3,
                    "end": 181.58,
                    "word": " where",
                    "probability": 0.90087890625
                },
                {
                    "start": 181.58,
                    "end": 181.68,
                    "word": " you",
                    "probability": 0.9013671875
                },
                {
                    "start": 181.68,
                    "end": 181.8,
                    "word": " have",
                    "probability": 0.8779296875
                },
                {
                    "start": 181.8,
                    "end": 181.92,
                    "word": " an",
                    "probability": 0.89501953125
                },
                {
                    "start": 181.92,
                    "end": 182.26,
                    "word": " input",
                    "probability": 0.82958984375
                },
                {
                    "start": 182.26,
                    "end": 182.8,
                    "word": " and",
                    "probability": 0.83203125
                },
                {
                    "start": 182.8,
                    "end": 183.6,
                    "word": " do",
                    "probability": 0.869140625
                },
                {
                    "start": 183.72,
                    "end": 183.98,
                    "word": " a prediction",
                    "probability": 0.83935546875
                },
                {
                    "start": 183.98,
                    "end": 184.36,
                    "word": " on",
                    "probability": 0.90576171875
                },
                {
                    "start": 184.36,
                    "end": 184.5,
                    "word": " the",
                    "probability": 0.7998046875
                },
                {
                    "start": 184.5,
                    "end": 184.86,
                    "word": " output",
                    "probability": 0.77978515625
                },
                {
                    "start": 184.86,
                    "end": 185.34,
                    "word": ".",
                    "probability": 0.712890625
                },
                {
                    "start": 185.74,
                    "end": 186.22,
                    "word": " In",
                    "probability": 0.59228515625
                },
                {
                    "start": 186.22,
                    "end": 186.46,
                    "word": " terms",
                    "probability": 0.3759765625
                },
                {
                    "start": 186.46,
                    "end": 186.62,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 186.62,
                    "end": 186.8,
                    "word": " this",
                    "probability": 0.82666015625
                },
                {
                    "start": 186.8,
                    "end": 186.84,
                    "word": ",",
                    "probability": 0.4443359375
                }
            ]
        },
        {
            "id": 29,
            "seek": 19242,
            "start": 186.84,
            "end": 190.12,
            "text": " it's the attack probability. And then you have the physical domain, which is the shutdown",
            "tokens": [
                51252,
                309,
                311,
                264,
                2690,
                8482,
                13,
                400,
                550,
                291,
                362,
                264,
                4001,
                9274,
                11,
                597,
                307,
                264,
                34927,
                51424
            ],
            "temperature": 0.0,
            "avg_logprob": -0.24885320663452148,
            "compression_ratio": 1.7709923664122138,
            "no_speech_prob": 0.0014324188232421875,
            "words": [
                {
                    "start": 186.84,
                    "end": 186.96,
                    "word": " it's",
                    "probability": 0.7705078125
                },
                {
                    "start": 186.96,
                    "end": 187.08,
                    "word": " the",
                    "probability": 0.7197265625
                },
                {
                    "start": 187.08,
                    "end": 187.3,
                    "word": " attack",
                    "probability": 0.75048828125
                },
                {
                    "start": 187.3,
                    "end": 187.66,
                    "word": " probability",
                    "probability": 0.81298828125
                },
                {
                    "start": 187.66,
                    "end": 188.02,
                    "word": ".",
                    "probability": 0.43115234375
                },
                {
                    "start": 188.02,
                    "end": 188.1,
                    "word": " And",
                    "probability": 0.7900390625
                },
                {
                    "start": 188.1,
                    "end": 188.36,
                    "word": " then",
                    "probability": 0.72216796875
                },
                {
                    "start": 188.36,
                    "end": 188.58,
                    "word": " you",
                    "probability": 0.5947265625
                },
                {
                    "start": 188.58,
                    "end": 188.68,
                    "word": " have",
                    "probability": 0.798828125
                },
                {
                    "start": 188.68,
                    "end": 188.8,
                    "word": " the",
                    "probability": 0.7998046875
                },
                {
                    "start": 188.8,
                    "end": 189.02,
                    "word": " physical",
                    "probability": 0.7265625
                },
                {
                    "start": 189.02,
                    "end": 189.38,
                    "word": " domain",
                    "probability": 0.87060546875
                },
                {
                    "start": 189.38,
                    "end": 189.48,
                    "word": ",",
                    "probability": 0.9521484375
                },
                {
                    "start": 189.48,
                    "end": 189.6,
                    "word": " which",
                    "probability": 0.91796875
                },
                {
                    "start": 189.6,
                    "end": 189.74,
                    "word": " is",
                    "probability": 0.90478515625
                },
                {
                    "start": 189.74,
                    "end": 189.86,
                    "word": " the",
                    "probability": 0.76611328125
                },
                {
                    "start": 189.86,
                    "end": 190.12,
                    "word": " shutdown",
                    "probability": 0.7744140625
                }
            ]
        },
        {
            "id": 30,
            "seek": 19242,
            "start": 190.12,
            "end": 195.36,
            "text": " infrastructure. For this talk, we'll focus on the machine learning model. In the next slide,",
            "tokens": [
                51424,
                6896,
                13,
                1171,
                341,
                751,
                11,
                321,
                603,
                1879,
                322,
                264,
                3479,
                2539,
                2316,
                13,
                682,
                264,
                958,
                4137,
                11,
                51688
            ],
            "temperature": 0.0,
            "avg_logprob": -0.24885320663452148,
            "compression_ratio": 1.7709923664122138,
            "no_speech_prob": 0.0014324188232421875,
            "words": [
                {
                    "start": 190.12,
                    "end": 190.64,
                    "word": " infrastructure",
                    "probability": 0.87841796875
                },
                {
                    "start": 190.64,
                    "end": 191.12,
                    "word": ".",
                    "probability": 0.974609375
                },
                {
                    "start": 191.26,
                    "end": 191.42,
                    "word": " For",
                    "probability": 0.88525390625
                },
                {
                    "start": 191.42,
                    "end": 191.62,
                    "word": " this",
                    "probability": 0.88671875
                },
                {
                    "start": 191.62,
                    "end": 191.84,
                    "word": " talk",
                    "probability": 0.7783203125
                },
                {
                    "start": 191.84,
                    "end": 191.86,
                    "word": ",",
                    "probability": 0.890625
                },
                {
                    "start": 191.86,
                    "end": 192.06,
                    "word": " we'll",
                    "probability": 0.9169921875
                },
                {
                    "start": 192.06,
                    "end": 192.32,
                    "word": " focus",
                    "probability": 0.84375
                },
                {
                    "start": 192.32,
                    "end": 192.48,
                    "word": " on",
                    "probability": 0.90673828125
                },
                {
                    "start": 192.48,
                    "end": 192.58,
                    "word": " the",
                    "probability": 0.44189453125
                },
                {
                    "start": 192.58,
                    "end": 192.84,
                    "word": " machine",
                    "probability": 0.72705078125
                },
                {
                    "start": 192.84,
                    "end": 193.12,
                    "word": " learning",
                    "probability": 0.83642578125
                },
                {
                    "start": 193.12,
                    "end": 193.48,
                    "word": " model",
                    "probability": 0.818359375
                },
                {
                    "start": 193.48,
                    "end": 193.96,
                    "word": ".",
                    "probability": 0.9921875
                },
                {
                    "start": 194.44,
                    "end": 194.58,
                    "word": " In",
                    "probability": 0.82666015625
                },
                {
                    "start": 194.58,
                    "end": 194.7,
                    "word": " the",
                    "probability": 0.80517578125
                },
                {
                    "start": 194.7,
                    "end": 194.86,
                    "word": " next",
                    "probability": 0.85693359375
                },
                {
                    "start": 194.86,
                    "end": 195.12,
                    "word": " slide",
                    "probability": 0.89794921875
                },
                {
                    "start": 195.12,
                    "end": 195.36,
                    "word": ",",
                    "probability": 0.9560546875
                }
            ]
        },
        {
            "id": 31,
            "seek": 22042,
            "start": 195.36,
            "end": 201.46,
            "text": " we will see our first type of attack, which is the linkage attack. With this attack, in the early",
            "tokens": [
                50368,
                321,
                486,
                536,
                527,
                700,
                2010,
                295,
                2690,
                11,
                597,
                307,
                264,
                49118,
                2690,
                13,
                2022,
                341,
                2690,
                11,
                294,
                264,
                2440,
                50668
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19810267125389405,
            "compression_ratio": 1.7798507462686568,
            "no_speech_prob": 0.0005030632019042969,
            "words": [
                {
                    "start": 195.36,
                    "end": 195.7,
                    "word": " we",
                    "probability": 0.5771484375
                },
                {
                    "start": 195.7,
                    "end": 195.94,
                    "word": " will",
                    "probability": 0.80517578125
                },
                {
                    "start": 195.94,
                    "end": 196.16,
                    "word": " see",
                    "probability": 0.82275390625
                },
                {
                    "start": 196.16,
                    "end": 196.38,
                    "word": " our",
                    "probability": 0.81640625
                },
                {
                    "start": 196.38,
                    "end": 196.82,
                    "word": " first",
                    "probability": 0.80615234375
                },
                {
                    "start": 196.82,
                    "end": 197.46,
                    "word": " type",
                    "probability": 0.927734375
                },
                {
                    "start": 197.46,
                    "end": 197.6,
                    "word": " of",
                    "probability": 0.90283203125
                },
                {
                    "start": 197.6,
                    "end": 197.82,
                    "word": " attack",
                    "probability": 0.7060546875
                },
                {
                    "start": 197.82,
                    "end": 197.94,
                    "word": ",",
                    "probability": 0.71435546875
                },
                {
                    "start": 197.94,
                    "end": 198.0,
                    "word": " which",
                    "probability": 0.91357421875
                },
                {
                    "start": 198.0,
                    "end": 198.12,
                    "word": " is",
                    "probability": 0.9052734375
                },
                {
                    "start": 198.12,
                    "end": 198.24,
                    "word": " the",
                    "probability": 0.8173828125
                },
                {
                    "start": 198.24,
                    "end": 198.42,
                    "word": " linkage",
                    "probability": 0.84814453125
                },
                {
                    "start": 198.42,
                    "end": 198.82,
                    "word": " attack",
                    "probability": 0.8388671875
                },
                {
                    "start": 198.82,
                    "end": 199.26,
                    "word": ".",
                    "probability": 0.88330078125
                },
                {
                    "start": 199.36,
                    "end": 199.74,
                    "word": " With",
                    "probability": 0.880859375
                },
                {
                    "start": 199.74,
                    "end": 199.94,
                    "word": " this",
                    "probability": 0.88525390625
                },
                {
                    "start": 199.94,
                    "end": 200.32,
                    "word": " attack",
                    "probability": 0.83740234375
                },
                {
                    "start": 200.32,
                    "end": 200.6,
                    "word": ",",
                    "probability": 0.372314453125
                },
                {
                    "start": 200.6,
                    "end": 201.08,
                    "word": " in",
                    "probability": 0.904296875
                },
                {
                    "start": 201.08,
                    "end": 201.28,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 201.28,
                    "end": 201.46,
                    "word": " early",
                    "probability": 0.9326171875
                }
            ]
        },
        {
            "id": 32,
            "seek": 22042,
            "start": 201.46,
            "end": 206.78,
            "text": " 90s, Tina Sweeney found that she was able to identify the identification of the governor of",
            "tokens": [
                50668,
                4289,
                82,
                11,
                28504,
                318,
                1259,
                2030,
                1352,
                300,
                750,
                390,
                1075,
                281,
                5876,
                264,
                22065,
                295,
                264,
                12965,
                295,
                50928
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19810267125389405,
            "compression_ratio": 1.7798507462686568,
            "no_speech_prob": 0.0005030632019042969,
            "words": [
                {
                    "start": 201.46,
                    "end": 202.08,
                    "word": " 90s",
                    "probability": 0.712646484375
                },
                {
                    "start": 202.08,
                    "end": 202.42,
                    "word": ",",
                    "probability": 0.92236328125
                },
                {
                    "start": 202.42,
                    "end": 202.64,
                    "word": " Tina",
                    "probability": 0.25732421875
                },
                {
                    "start": 202.64,
                    "end": 202.88,
                    "word": " Sweeney",
                    "probability": 0.8898111979166666
                },
                {
                    "start": 202.88,
                    "end": 203.18,
                    "word": " found",
                    "probability": 0.87646484375
                },
                {
                    "start": 203.18,
                    "end": 203.42,
                    "word": " that",
                    "probability": 0.8046875
                },
                {
                    "start": 203.42,
                    "end": 203.66,
                    "word": " she",
                    "probability": 0.79931640625
                },
                {
                    "start": 203.66,
                    "end": 203.74,
                    "word": " was",
                    "probability": 0.89697265625
                },
                {
                    "start": 203.74,
                    "end": 203.9,
                    "word": " able",
                    "probability": 0.900390625
                },
                {
                    "start": 203.9,
                    "end": 204.06,
                    "word": " to",
                    "probability": 0.90234375
                },
                {
                    "start": 204.06,
                    "end": 204.54,
                    "word": " identify",
                    "probability": 0.90185546875
                },
                {
                    "start": 204.54,
                    "end": 205.4,
                    "word": " the",
                    "probability": 0.802734375
                },
                {
                    "start": 205.4,
                    "end": 205.88,
                    "word": " identification",
                    "probability": 0.884765625
                },
                {
                    "start": 205.88,
                    "end": 206.32,
                    "word": " of",
                    "probability": 0.9091796875
                },
                {
                    "start": 206.32,
                    "end": 206.44,
                    "word": " the",
                    "probability": 0.8095703125
                },
                {
                    "start": 206.44,
                    "end": 206.6,
                    "word": " governor",
                    "probability": 0.63671875
                },
                {
                    "start": 206.6,
                    "end": 206.78,
                    "word": " of",
                    "probability": 0.900390625
                }
            ]
        },
        {
            "id": 33,
            "seek": 22042,
            "start": 206.78,
            "end": 211.52,
            "text": " Massachusetts by linking his health records to a voter information just by using three variables,",
            "tokens": [
                50928,
                19979,
                538,
                25775,
                702,
                1585,
                7724,
                281,
                257,
                21722,
                1589,
                445,
                538,
                1228,
                1045,
                9102,
                11,
                51168
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19810267125389405,
            "compression_ratio": 1.7798507462686568,
            "no_speech_prob": 0.0005030632019042969,
            "words": [
                {
                    "start": 206.78,
                    "end": 207.18,
                    "word": " Massachusetts",
                    "probability": 0.67333984375
                },
                {
                    "start": 207.18,
                    "end": 207.6,
                    "word": " by",
                    "probability": 0.86474609375
                },
                {
                    "start": 207.6,
                    "end": 207.86,
                    "word": " linking",
                    "probability": 0.7568359375
                },
                {
                    "start": 207.86,
                    "end": 208.06,
                    "word": " his",
                    "probability": 0.90087890625
                },
                {
                    "start": 208.06,
                    "end": 208.28,
                    "word": " health",
                    "probability": 0.89306640625
                },
                {
                    "start": 208.28,
                    "end": 208.7,
                    "word": " records",
                    "probability": 0.7822265625
                },
                {
                    "start": 208.7,
                    "end": 208.92,
                    "word": " to",
                    "probability": 0.8994140625
                },
                {
                    "start": 209.06,
                    "end": 209.22,
                    "word": " a voter",
                    "probability": 0.8369140625
                },
                {
                    "start": 209.22,
                    "end": 209.78,
                    "word": " information",
                    "probability": 0.7216796875
                },
                {
                    "start": 209.78,
                    "end": 210.06,
                    "word": " just",
                    "probability": 0.5546875
                },
                {
                    "start": 210.06,
                    "end": 210.24,
                    "word": " by",
                    "probability": 0.8935546875
                },
                {
                    "start": 210.24,
                    "end": 210.5,
                    "word": " using",
                    "probability": 0.8271484375
                },
                {
                    "start": 210.5,
                    "end": 210.84,
                    "word": " three",
                    "probability": 0.8466796875
                },
                {
                    "start": 210.84,
                    "end": 211.32,
                    "word": " variables",
                    "probability": 0.8291015625
                },
                {
                    "start": 211.32,
                    "end": 211.52,
                    "word": ",",
                    "probability": 0.71240234375
                }
            ]
        },
        {
            "id": 34,
            "seek": 22042,
            "start": 211.52,
            "end": 217.56,
            "text": " his gender, date of birth, and a zip code. In the next slide, she was able to identify that",
            "tokens": [
                51168,
                702,
                7898,
                11,
                4002,
                295,
                3965,
                11,
                293,
                257,
                20730,
                3089,
                13,
                682,
                264,
                958,
                4137,
                11,
                750,
                390,
                1075,
                281,
                5876,
                300,
                51468
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19810267125389405,
            "compression_ratio": 1.7798507462686568,
            "no_speech_prob": 0.0005030632019042969,
            "words": [
                {
                    "start": 211.52,
                    "end": 211.74,
                    "word": " his",
                    "probability": 0.892578125
                },
                {
                    "start": 211.74,
                    "end": 212.14,
                    "word": " gender",
                    "probability": 0.7998046875
                },
                {
                    "start": 212.14,
                    "end": 212.7,
                    "word": ",",
                    "probability": 0.9853515625
                },
                {
                    "start": 212.7,
                    "end": 212.8,
                    "word": " date",
                    "probability": 0.8857421875
                },
                {
                    "start": 212.8,
                    "end": 212.98,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 212.98,
                    "end": 213.26,
                    "word": " birth",
                    "probability": 0.9150390625
                },
                {
                    "start": 213.26,
                    "end": 213.44,
                    "word": ",",
                    "probability": 0.65380859375
                },
                {
                    "start": 213.44,
                    "end": 213.62,
                    "word": " and",
                    "probability": 0.896484375
                },
                {
                    "start": 213.76,
                    "end": 213.92,
                    "word": " a zip",
                    "probability": 0.79833984375
                },
                {
                    "start": 213.92,
                    "end": 214.2,
                    "word": " code",
                    "probability": 0.82080078125
                },
                {
                    "start": 214.2,
                    "end": 214.64,
                    "word": ".",
                    "probability": 0.994140625
                },
                {
                    "start": 214.82,
                    "end": 214.92,
                    "word": " In",
                    "probability": 0.86279296875
                },
                {
                    "start": 214.92,
                    "end": 215.04,
                    "word": " the",
                    "probability": 0.81396484375
                },
                {
                    "start": 215.04,
                    "end": 215.24,
                    "word": " next",
                    "probability": 0.859375
                },
                {
                    "start": 215.24,
                    "end": 215.6,
                    "word": " slide",
                    "probability": 0.89453125
                },
                {
                    "start": 215.6,
                    "end": 216.14,
                    "word": ",",
                    "probability": 0.95751953125
                },
                {
                    "start": 216.14,
                    "end": 216.3,
                    "word": " she",
                    "probability": 0.82373046875
                },
                {
                    "start": 216.3,
                    "end": 216.42,
                    "word": " was",
                    "probability": 0.900390625
                },
                {
                    "start": 216.42,
                    "end": 216.62,
                    "word": " able",
                    "probability": 0.89404296875
                },
                {
                    "start": 216.62,
                    "end": 216.76,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 216.76,
                    "end": 217.14,
                    "word": " identify",
                    "probability": 0.89306640625
                },
                {
                    "start": 217.14,
                    "end": 217.56,
                    "word": " that",
                    "probability": 0.8681640625
                }
            ]
        },
        {
            "id": 35,
            "seek": 22042,
            "start": 218.1,
            "end": 223.36,
            "text": " using these three variables alone, she was able to uniquely identify 87% of the population in the",
            "tokens": [
                51508,
                1228,
                613,
                1045,
                9102,
                3312,
                11,
                750,
                390,
                1075,
                281,
                31474,
                5876,
                27990,
                4,
                295,
                264,
                4415,
                294,
                264,
                51760
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19810267125389405,
            "compression_ratio": 1.7798507462686568,
            "no_speech_prob": 0.0005030632019042969,
            "words": [
                {
                    "start": 218.1,
                    "end": 218.54,
                    "word": " using",
                    "probability": 0.67138671875
                },
                {
                    "start": 218.54,
                    "end": 218.82,
                    "word": " these",
                    "probability": 0.7119140625
                },
                {
                    "start": 218.82,
                    "end": 219.0,
                    "word": " three",
                    "probability": 0.81689453125
                },
                {
                    "start": 219.0,
                    "end": 219.34,
                    "word": " variables",
                    "probability": 0.8427734375
                },
                {
                    "start": 219.34,
                    "end": 219.72,
                    "word": " alone",
                    "probability": 0.91015625
                },
                {
                    "start": 219.72,
                    "end": 219.86,
                    "word": ",",
                    "probability": 0.85302734375
                },
                {
                    "start": 219.86,
                    "end": 219.94,
                    "word": " she",
                    "probability": 0.82080078125
                },
                {
                    "start": 219.94,
                    "end": 220.06,
                    "word": " was",
                    "probability": 0.89990234375
                },
                {
                    "start": 220.06,
                    "end": 220.22,
                    "word": " able",
                    "probability": 0.8955078125
                },
                {
                    "start": 220.22,
                    "end": 220.44,
                    "word": " to",
                    "probability": 0.91455078125
                },
                {
                    "start": 220.44,
                    "end": 220.86,
                    "word": " uniquely",
                    "probability": 0.87353515625
                },
                {
                    "start": 220.86,
                    "end": 221.32,
                    "word": " identify",
                    "probability": 0.880859375
                },
                {
                    "start": 221.32,
                    "end": 221.9,
                    "word": " 87",
                    "probability": 0.87353515625
                },
                {
                    "start": 221.9,
                    "end": 222.24,
                    "word": "%",
                    "probability": 0.9189453125
                },
                {
                    "start": 222.24,
                    "end": 222.42,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 222.42,
                    "end": 222.54,
                    "word": " the",
                    "probability": 0.8173828125
                },
                {
                    "start": 222.54,
                    "end": 222.96,
                    "word": " population",
                    "probability": 0.8701171875
                },
                {
                    "start": 222.96,
                    "end": 223.22,
                    "word": " in",
                    "probability": 0.9052734375
                },
                {
                    "start": 223.22,
                    "end": 223.36,
                    "word": " the",
                    "probability": 0.81591796875
                }
            ]
        },
        {
            "id": 36,
            "seek": 24846,
            "start": 223.36,
            "end": 231.6,
            "text": " U.S. using the census, the 1990 census. In 2000, Philip Goh identified that using these three",
            "tokens": [
                50364,
                624,
                13,
                50,
                13,
                1228,
                264,
                23725,
                11,
                264,
                13384,
                23725,
                13,
                682,
                8132,
                11,
                21144,
                1037,
                71,
                9234,
                300,
                1228,
                613,
                1045,
                50776
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22033006153749615,
            "compression_ratio": 1.524390243902439,
            "no_speech_prob": 0.0004582405090332031,
            "words": [
                {
                    "start": 223.36,
                    "end": 223.42,
                    "word": " U",
                    "probability": 0.00887298583984375
                },
                {
                    "start": 223.42,
                    "end": 223.74,
                    "word": ".S",
                    "probability": 0.984619140625
                },
                {
                    "start": 223.74,
                    "end": 224.4,
                    "word": ".",
                    "probability": 0.87646484375
                },
                {
                    "start": 224.5,
                    "end": 224.74,
                    "word": " using",
                    "probability": 0.66650390625
                },
                {
                    "start": 224.74,
                    "end": 225.06,
                    "word": " the",
                    "probability": 0.810546875
                },
                {
                    "start": 225.06,
                    "end": 225.44,
                    "word": " census",
                    "probability": 0.8076171875
                },
                {
                    "start": 225.44,
                    "end": 226.4,
                    "word": ",",
                    "probability": 0.7685546875
                },
                {
                    "start": 226.4,
                    "end": 226.5,
                    "word": " the",
                    "probability": 0.80419921875
                },
                {
                    "start": 226.5,
                    "end": 226.8,
                    "word": " 1990",
                    "probability": 0.77001953125
                },
                {
                    "start": 226.8,
                    "end": 227.16,
                    "word": " census",
                    "probability": 0.77099609375
                },
                {
                    "start": 227.16,
                    "end": 227.6,
                    "word": ".",
                    "probability": 0.943359375
                },
                {
                    "start": 227.6,
                    "end": 227.78,
                    "word": " In",
                    "probability": 0.89794921875
                },
                {
                    "start": 227.78,
                    "end": 228.92,
                    "word": " 2000",
                    "probability": 0.82568359375
                },
                {
                    "start": 228.92,
                    "end": 229.22,
                    "word": ",",
                    "probability": 0.83935546875
                },
                {
                    "start": 229.22,
                    "end": 229.5,
                    "word": " Philip",
                    "probability": 0.48583984375
                },
                {
                    "start": 229.5,
                    "end": 229.9,
                    "word": " Goh",
                    "probability": 0.6014404296875
                },
                {
                    "start": 229.9,
                    "end": 230.42,
                    "word": " identified",
                    "probability": 0.87646484375
                },
                {
                    "start": 230.42,
                    "end": 230.8,
                    "word": " that",
                    "probability": 0.86083984375
                },
                {
                    "start": 230.8,
                    "end": 231.12,
                    "word": " using",
                    "probability": 0.734375
                },
                {
                    "start": 231.12,
                    "end": 231.42,
                    "word": " these",
                    "probability": 0.7177734375
                },
                {
                    "start": 231.42,
                    "end": 231.6,
                    "word": " three",
                    "probability": 0.84326171875
                }
            ]
        },
        {
            "id": 37,
            "seek": 24846,
            "start": 231.6,
            "end": 238.74,
            "text": " records, he was able to identify only 64% of the population. This could be a result of people",
            "tokens": [
                50776,
                7724,
                11,
                415,
                390,
                1075,
                281,
                5876,
                787,
                12145,
                4,
                295,
                264,
                4415,
                13,
                639,
                727,
                312,
                257,
                1874,
                295,
                561,
                51136
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22033006153749615,
            "compression_ratio": 1.524390243902439,
            "no_speech_prob": 0.0004582405090332031,
            "words": [
                {
                    "start": 231.6,
                    "end": 232.08,
                    "word": " records",
                    "probability": 0.75244140625
                },
                {
                    "start": 232.08,
                    "end": 232.74,
                    "word": ",",
                    "probability": 0.6708984375
                },
                {
                    "start": 232.74,
                    "end": 232.84,
                    "word": " he",
                    "probability": 0.88134765625
                },
                {
                    "start": 232.84,
                    "end": 232.96,
                    "word": " was",
                    "probability": 0.8974609375
                },
                {
                    "start": 232.96,
                    "end": 233.22,
                    "word": " able",
                    "probability": 0.884765625
                },
                {
                    "start": 233.22,
                    "end": 233.9,
                    "word": " to",
                    "probability": 0.90478515625
                },
                {
                    "start": 233.9,
                    "end": 234.4,
                    "word": " identify",
                    "probability": 0.64404296875
                },
                {
                    "start": 234.4,
                    "end": 234.66,
                    "word": " only",
                    "probability": 0.81005859375
                },
                {
                    "start": 234.66,
                    "end": 235.1,
                    "word": " 64",
                    "probability": 0.87744140625
                },
                {
                    "start": 235.1,
                    "end": 235.42,
                    "word": "%",
                    "probability": 0.8359375
                },
                {
                    "start": 235.42,
                    "end": 235.56,
                    "word": " of",
                    "probability": 0.90087890625
                },
                {
                    "start": 235.56,
                    "end": 235.64,
                    "word": " the",
                    "probability": 0.81591796875
                },
                {
                    "start": 235.64,
                    "end": 236.12,
                    "word": " population",
                    "probability": 0.87890625
                },
                {
                    "start": 236.12,
                    "end": 236.64,
                    "word": ".",
                    "probability": 0.97900390625
                },
                {
                    "start": 236.64,
                    "end": 236.8,
                    "word": " This",
                    "probability": 0.81201171875
                },
                {
                    "start": 236.8,
                    "end": 236.94,
                    "word": " could",
                    "probability": 0.81640625
                },
                {
                    "start": 236.94,
                    "end": 237.34,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 237.54,
                    "end": 237.78,
                    "word": " a result",
                    "probability": 0.8310546875
                },
                {
                    "start": 237.78,
                    "end": 238.26,
                    "word": " of",
                    "probability": 0.90869140625
                },
                {
                    "start": 238.26,
                    "end": 238.74,
                    "word": " people",
                    "probability": 0.9267578125
                }
            ]
        },
        {
            "id": 38,
            "seek": 24846,
            "start": 238.74,
            "end": 245.84,
            "text": " urbanizing more, moving from the rural areas to the cities. But this provides an important point",
            "tokens": [
                51136,
                9681,
                3319,
                544,
                11,
                2684,
                490,
                264,
                11165,
                3179,
                281,
                264,
                6486,
                13,
                583,
                341,
                6417,
                364,
                1021,
                935,
                51492
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22033006153749615,
            "compression_ratio": 1.524390243902439,
            "no_speech_prob": 0.0004582405090332031,
            "words": [
                {
                    "start": 238.74,
                    "end": 239.38,
                    "word": " urbanizing",
                    "probability": 0.79541015625
                },
                {
                    "start": 239.38,
                    "end": 239.66,
                    "word": " more",
                    "probability": 0.87255859375
                },
                {
                    "start": 239.66,
                    "end": 239.74,
                    "word": ",",
                    "probability": 0.7666015625
                },
                {
                    "start": 239.74,
                    "end": 239.94,
                    "word": " moving",
                    "probability": 0.85107421875
                },
                {
                    "start": 239.94,
                    "end": 240.16,
                    "word": " from",
                    "probability": 0.80322265625
                },
                {
                    "start": 240.16,
                    "end": 240.3,
                    "word": " the",
                    "probability": 0.77392578125
                },
                {
                    "start": 240.3,
                    "end": 240.46,
                    "word": " rural",
                    "probability": 0.8671875
                },
                {
                    "start": 240.46,
                    "end": 240.92,
                    "word": " areas",
                    "probability": 0.87646484375
                },
                {
                    "start": 240.92,
                    "end": 241.56,
                    "word": " to",
                    "probability": 0.8916015625
                },
                {
                    "start": 241.56,
                    "end": 242.44,
                    "word": " the",
                    "probability": 0.77294921875
                },
                {
                    "start": 242.44,
                    "end": 242.78,
                    "word": " cities",
                    "probability": 0.7265625
                },
                {
                    "start": 242.78,
                    "end": 243.16,
                    "word": ".",
                    "probability": 0.6767578125
                },
                {
                    "start": 243.16,
                    "end": 243.56,
                    "word": " But",
                    "probability": 0.8720703125
                },
                {
                    "start": 243.56,
                    "end": 244.1,
                    "word": " this",
                    "probability": 0.779296875
                },
                {
                    "start": 244.1,
                    "end": 245.08,
                    "word": " provides",
                    "probability": 0.8125
                },
                {
                    "start": 245.08,
                    "end": 245.24,
                    "word": " an",
                    "probability": 0.83740234375
                },
                {
                    "start": 245.24,
                    "end": 245.48,
                    "word": " important",
                    "probability": 0.7783203125
                },
                {
                    "start": 245.48,
                    "end": 245.84,
                    "word": " point",
                    "probability": 0.8603515625
                }
            ]
        },
        {
            "id": 39,
            "seek": 24846,
            "start": 245.84,
            "end": 251.4,
            "text": " where using just very simple features, attackers can identify back to the original source.",
            "tokens": [
                51492,
                689,
                1228,
                445,
                588,
                2199,
                4122,
                11,
                45129,
                393,
                5876,
                646,
                281,
                264,
                3380,
                4009,
                13,
                51732
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22033006153749615,
            "compression_ratio": 1.524390243902439,
            "no_speech_prob": 0.0004582405090332031,
            "words": [
                {
                    "start": 245.84,
                    "end": 246.26,
                    "word": " where",
                    "probability": 0.8125
                },
                {
                    "start": 246.26,
                    "end": 246.88,
                    "word": " using",
                    "probability": 0.55322265625
                },
                {
                    "start": 246.88,
                    "end": 247.1,
                    "word": " just",
                    "probability": 0.86962890625
                },
                {
                    "start": 247.1,
                    "end": 247.32,
                    "word": " very",
                    "probability": 0.69677734375
                },
                {
                    "start": 247.32,
                    "end": 247.62,
                    "word": " simple",
                    "probability": 0.908203125
                },
                {
                    "start": 247.62,
                    "end": 248.06,
                    "word": " features",
                    "probability": 0.7666015625
                },
                {
                    "start": 248.06,
                    "end": 248.24,
                    "word": ",",
                    "probability": 0.79150390625
                },
                {
                    "start": 248.24,
                    "end": 248.58,
                    "word": " attackers",
                    "probability": 0.86474609375
                },
                {
                    "start": 248.58,
                    "end": 248.96,
                    "word": " can",
                    "probability": 0.9052734375
                },
                {
                    "start": 248.96,
                    "end": 249.32,
                    "word": " identify",
                    "probability": 0.8642578125
                },
                {
                    "start": 249.32,
                    "end": 249.76,
                    "word": " back",
                    "probability": 0.744140625
                },
                {
                    "start": 249.76,
                    "end": 249.92,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 249.92,
                    "end": 250.04,
                    "word": " the",
                    "probability": 0.806640625
                },
                {
                    "start": 250.04,
                    "end": 250.26,
                    "word": " original",
                    "probability": 0.84228515625
                },
                {
                    "start": 250.26,
                    "end": 250.74,
                    "word": " source",
                    "probability": 0.71630859375
                },
                {
                    "start": 250.74,
                    "end": 251.4,
                    "word": ".",
                    "probability": 0.99365234375
                }
            ]
        },
        {
            "id": 40,
            "seek": 27706,
            "start": 251.93,
            "end": 258.52,
            "text": " In the next slide, we look at another popular linkage attack. In 2006, Netflix released a",
            "tokens": [
                50392,
                682,
                264,
                958,
                4137,
                11,
                321,
                574,
                412,
                1071,
                3743,
                49118,
                2690,
                13,
                682,
                14062,
                11,
                12778,
                4736,
                257,
                50704
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21704545562917535,
            "compression_ratio": 1.6308243727598566,
            "no_speech_prob": 9.167194366455078e-05,
            "words": [
                {
                    "start": 251.93,
                    "end": 252.18,
                    "word": " In",
                    "probability": 0.67822265625
                },
                {
                    "start": 252.18,
                    "end": 252.28,
                    "word": " the",
                    "probability": 0.8046875
                },
                {
                    "start": 252.28,
                    "end": 252.48,
                    "word": " next",
                    "probability": 0.86083984375
                },
                {
                    "start": 252.48,
                    "end": 252.74,
                    "word": " slide",
                    "probability": 0.87890625
                },
                {
                    "start": 252.74,
                    "end": 252.9,
                    "word": ",",
                    "probability": 0.7607421875
                },
                {
                    "start": 252.9,
                    "end": 252.98,
                    "word": " we",
                    "probability": 0.85302734375
                },
                {
                    "start": 252.98,
                    "end": 253.56,
                    "word": " look",
                    "probability": 0.78173828125
                },
                {
                    "start": 253.56,
                    "end": 253.7,
                    "word": " at",
                    "probability": 0.91064453125
                },
                {
                    "start": 253.7,
                    "end": 253.98,
                    "word": " another",
                    "probability": 0.8134765625
                },
                {
                    "start": 253.98,
                    "end": 254.6,
                    "word": " popular",
                    "probability": 0.904296875
                },
                {
                    "start": 254.6,
                    "end": 255.0,
                    "word": " linkage",
                    "probability": 0.8740234375
                },
                {
                    "start": 255.0,
                    "end": 255.48,
                    "word": " attack",
                    "probability": 0.84765625
                },
                {
                    "start": 255.48,
                    "end": 255.98,
                    "word": ".",
                    "probability": 0.89990234375
                },
                {
                    "start": 256.38,
                    "end": 256.64,
                    "word": " In",
                    "probability": 0.8984375
                },
                {
                    "start": 256.64,
                    "end": 257.3,
                    "word": " 2006",
                    "probability": 0.8564453125
                },
                {
                    "start": 257.3,
                    "end": 257.44,
                    "word": ",",
                    "probability": 0.89404296875
                },
                {
                    "start": 257.44,
                    "end": 257.66,
                    "word": " Netflix",
                    "probability": 0.853515625
                },
                {
                    "start": 257.66,
                    "end": 258.16,
                    "word": " released",
                    "probability": 0.84912109375
                },
                {
                    "start": 258.32,
                    "end": 258.52,
                    "word": " a public",
                    "probability": 0.86572265625
                }
            ]
        },
        {
            "id": 41,
            "seek": 27706,
            "start": 258.52,
            "end": 263.3,
            "text": " public data set, a competition where they're offering a prize to improve their recommendation.",
            "tokens": [
                50704,
                1908,
                1412,
                992,
                11,
                257,
                6211,
                689,
                436,
                434,
                8745,
                257,
                12818,
                281,
                3470,
                641,
                11879,
                13,
                50900
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21704545562917535,
            "compression_ratio": 1.6308243727598566,
            "no_speech_prob": 9.167194366455078e-05,
            "words": [
                {
                    "start": 258.52,
                    "end": 258.76,
                    "word": " data",
                    "probability": 0.44482421875
                },
                {
                    "start": 258.76,
                    "end": 259.12,
                    "word": " set",
                    "probability": 0.4453125
                },
                {
                    "start": 259.12,
                    "end": 259.22,
                    "word": ",",
                    "probability": 0.7431640625
                },
                {
                    "start": 259.34,
                    "end": 259.82,
                    "word": " a competition",
                    "probability": 0.86279296875
                },
                {
                    "start": 259.82,
                    "end": 260.22,
                    "word": " where",
                    "probability": 0.82080078125
                },
                {
                    "start": 260.22,
                    "end": 260.4,
                    "word": " they're",
                    "probability": 0.72607421875
                },
                {
                    "start": 260.4,
                    "end": 260.62,
                    "word": " offering",
                    "probability": 0.7353515625
                },
                {
                    "start": 260.78,
                    "end": 261.02,
                    "word": " a prize",
                    "probability": 0.583984375
                },
                {
                    "start": 261.02,
                    "end": 261.2,
                    "word": " to",
                    "probability": 0.90380859375
                },
                {
                    "start": 261.2,
                    "end": 261.44,
                    "word": " improve",
                    "probability": 0.6748046875
                },
                {
                    "start": 261.44,
                    "end": 261.58,
                    "word": " their",
                    "probability": 0.572265625
                },
                {
                    "start": 261.58,
                    "end": 262.06,
                    "word": " recommendation",
                    "probability": 0.62060546875
                },
                {
                    "start": 262.06,
                    "end": 262.56,
                    "word": ".",
                    "probability": 0.87841796875
                },
                {
                    "start": 263.08,
                    "end": 263.3,
                    "word": " And",
                    "probability": 0.498779296875
                }
            ]
        },
        {
            "id": 42,
            "seek": 27706,
            "start": 263.3,
            "end": 269.56,
            "text": " And their FAQs denoted that there is no private information in the data set itself,",
            "tokens": [
                50940,
                400,
                641,
                19894,
                48,
                82,
                1441,
                23325,
                300,
                456,
                307,
                572,
                4551,
                1589,
                294,
                264,
                1412,
                992,
                2564,
                11,
                51256
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21704545562917535,
            "compression_ratio": 1.6308243727598566,
            "no_speech_prob": 9.167194366455078e-05,
            "words": [
                {
                    "start": 263.3,
                    "end": 263.7,
                    "word": " their",
                    "probability": 0.81982421875
                },
                {
                    "start": 263.7,
                    "end": 264.9,
                    "word": " FAQs",
                    "probability": 0.88671875
                },
                {
                    "start": 264.9,
                    "end": 265.18,
                    "word": " denoted",
                    "probability": 0.810791015625
                },
                {
                    "start": 265.18,
                    "end": 265.54,
                    "word": " that",
                    "probability": 0.875
                },
                {
                    "start": 265.54,
                    "end": 266.0,
                    "word": " there",
                    "probability": 0.81884765625
                },
                {
                    "start": 266.0,
                    "end": 266.2,
                    "word": " is",
                    "probability": 0.892578125
                },
                {
                    "start": 266.2,
                    "end": 266.44,
                    "word": " no",
                    "probability": 0.91357421875
                },
                {
                    "start": 266.44,
                    "end": 266.72,
                    "word": " private",
                    "probability": 0.80078125
                },
                {
                    "start": 266.72,
                    "end": 267.32,
                    "word": " information",
                    "probability": 0.7451171875
                },
                {
                    "start": 267.32,
                    "end": 268.02,
                    "word": " in",
                    "probability": 0.88623046875
                },
                {
                    "start": 268.02,
                    "end": 268.12,
                    "word": " the",
                    "probability": 0.415771484375
                },
                {
                    "start": 268.12,
                    "end": 268.32,
                    "word": " data",
                    "probability": 0.69580078125
                },
                {
                    "start": 268.32,
                    "end": 268.46,
                    "word": " set",
                    "probability": 0.830078125
                },
                {
                    "start": 268.46,
                    "end": 268.9,
                    "word": " itself",
                    "probability": 0.78955078125
                },
                {
                    "start": 268.9,
                    "end": 269.38,
                    "word": ",",
                    "probability": 0.57080078125
                },
                {
                    "start": 269.38,
                    "end": 269.56,
                    "word": " but",
                    "probability": 0.89111328125
                }
            ]
        },
        {
            "id": 43,
            "seek": 27706,
            "start": 269.56,
            "end": 275.12,
            "text": " but researchers have found that if they were able to link the Netflix accounts back to the IMDb,",
            "tokens": [
                51256,
                457,
                10309,
                362,
                1352,
                300,
                498,
                436,
                645,
                1075,
                281,
                2113,
                264,
                12778,
                9402,
                646,
                281,
                264,
                21463,
                35,
                65,
                11,
                51532
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21704545562917535,
            "compression_ratio": 1.6308243727598566,
            "no_speech_prob": 9.167194366455078e-05,
            "words": [
                {
                    "start": 269.56,
                    "end": 269.94,
                    "word": " researchers",
                    "probability": 0.80224609375
                },
                {
                    "start": 269.94,
                    "end": 270.16,
                    "word": " have",
                    "probability": 0.876953125
                },
                {
                    "start": 270.16,
                    "end": 270.44,
                    "word": " found",
                    "probability": 0.8759765625
                },
                {
                    "start": 270.44,
                    "end": 270.6,
                    "word": " that",
                    "probability": 0.8642578125
                },
                {
                    "start": 270.6,
                    "end": 270.72,
                    "word": " if",
                    "probability": 0.89111328125
                },
                {
                    "start": 270.72,
                    "end": 270.82,
                    "word": " they",
                    "probability": 0.7236328125
                },
                {
                    "start": 270.82,
                    "end": 270.92,
                    "word": " were",
                    "probability": 0.833984375
                },
                {
                    "start": 270.92,
                    "end": 271.2,
                    "word": " able",
                    "probability": 0.89697265625
                },
                {
                    "start": 271.2,
                    "end": 271.38,
                    "word": " to",
                    "probability": 0.908203125
                },
                {
                    "start": 271.38,
                    "end": 271.6,
                    "word": " link",
                    "probability": 0.88525390625
                },
                {
                    "start": 271.6,
                    "end": 271.74,
                    "word": " the",
                    "probability": 0.759765625
                },
                {
                    "start": 271.74,
                    "end": 272.2,
                    "word": " Netflix",
                    "probability": 0.8359375
                },
                {
                    "start": 272.2,
                    "end": 272.88,
                    "word": " accounts",
                    "probability": 0.7236328125
                },
                {
                    "start": 272.88,
                    "end": 273.42,
                    "word": " back",
                    "probability": 0.80615234375
                },
                {
                    "start": 273.42,
                    "end": 273.68,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 273.68,
                    "end": 273.82,
                    "word": " the",
                    "probability": 0.6943359375
                },
                {
                    "start": 273.82,
                    "end": 274.44,
                    "word": " IMDb",
                    "probability": 0.8260091145833334
                },
                {
                    "start": 274.44,
                    "end": 274.92,
                    "word": ",",
                    "probability": 0.66162109375
                },
                {
                    "start": 274.92,
                    "end": 275.12,
                    "word": " just",
                    "probability": 0.86376953125
                }
            ]
        },
        {
            "id": 44,
            "seek": 27706,
            "start": 275.12,
            "end": 280.0,
            "text": " just by using the name of the person on the IMDb, their public dates, and public ratings,",
            "tokens": [
                51532,
                445,
                538,
                1228,
                264,
                1315,
                295,
                264,
                954,
                322,
                264,
                21463,
                35,
                65,
                11,
                641,
                1908,
                11691,
                11,
                293,
                1908,
                24603,
                11,
                51784
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21704545562917535,
            "compression_ratio": 1.6308243727598566,
            "no_speech_prob": 9.167194366455078e-05,
            "words": [
                {
                    "start": 275.12,
                    "end": 275.26,
                    "word": " by",
                    "probability": 0.900390625
                },
                {
                    "start": 275.26,
                    "end": 275.52,
                    "word": " using",
                    "probability": 0.83740234375
                },
                {
                    "start": 275.52,
                    "end": 275.76,
                    "word": " the",
                    "probability": 0.8115234375
                },
                {
                    "start": 275.76,
                    "end": 276.08,
                    "word": " name",
                    "probability": 0.8037109375
                },
                {
                    "start": 276.08,
                    "end": 276.36,
                    "word": " of",
                    "probability": 0.90185546875
                },
                {
                    "start": 276.36,
                    "end": 276.46,
                    "word": " the",
                    "probability": 0.72021484375
                },
                {
                    "start": 276.46,
                    "end": 276.64,
                    "word": " person",
                    "probability": 0.7646484375
                },
                {
                    "start": 276.64,
                    "end": 276.78,
                    "word": " on",
                    "probability": 0.82275390625
                },
                {
                    "start": 276.78,
                    "end": 276.86,
                    "word": " the",
                    "probability": 0.364501953125
                },
                {
                    "start": 276.86,
                    "end": 277.42,
                    "word": " IMDb",
                    "probability": 0.9200846354166666
                },
                {
                    "start": 277.42,
                    "end": 277.7,
                    "word": ",",
                    "probability": 0.9736328125
                },
                {
                    "start": 277.7,
                    "end": 277.82,
                    "word": " their",
                    "probability": 0.87109375
                },
                {
                    "start": 277.82,
                    "end": 278.14,
                    "word": " public",
                    "probability": 0.88818359375
                },
                {
                    "start": 278.14,
                    "end": 278.6,
                    "word": " dates",
                    "probability": 0.80859375
                },
                {
                    "start": 278.6,
                    "end": 278.68,
                    "word": ",",
                    "probability": 0.33740234375
                },
                {
                    "start": 278.68,
                    "end": 278.9,
                    "word": " and",
                    "probability": 0.8916015625
                },
                {
                    "start": 278.9,
                    "end": 279.18,
                    "word": " public",
                    "probability": 0.8701171875
                },
                {
                    "start": 279.18,
                    "end": 279.62,
                    "word": " ratings",
                    "probability": 0.72509765625
                },
                {
                    "start": 279.62,
                    "end": 280.0,
                    "word": ",",
                    "probability": 0.94482421875
                }
            ]
        },
        {
            "id": 45,
            "seek": 30660,
            "start": 280.0,
            "end": 287.82,
            "text": " they'll identify personable information back to Netflix. You might be wondering why this might be",
            "tokens": [
                50364,
                436,
                603,
                5876,
                954,
                712,
                1589,
                646,
                281,
                12778,
                13,
                509,
                1062,
                312,
                6359,
                983,
                341,
                1062,
                312,
                50756
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23506943997409607,
            "compression_ratio": 1.7142857142857142,
            "no_speech_prob": 0.00045108795166015625,
            "words": [
                {
                    "start": 280.0,
                    "end": 280.14,
                    "word": " they'll",
                    "probability": 0.15081787109375
                },
                {
                    "start": 280.14,
                    "end": 280.58,
                    "word": " identify",
                    "probability": 0.82958984375
                },
                {
                    "start": 280.58,
                    "end": 282.02,
                    "word": " personable",
                    "probability": 0.830078125
                },
                {
                    "start": 282.02,
                    "end": 283.4,
                    "word": " information",
                    "probability": 0.69287109375
                },
                {
                    "start": 283.4,
                    "end": 283.78,
                    "word": " back",
                    "probability": 0.80810546875
                },
                {
                    "start": 283.78,
                    "end": 284.0,
                    "word": " to",
                    "probability": 0.90380859375
                },
                {
                    "start": 284.0,
                    "end": 284.32,
                    "word": " Netflix",
                    "probability": 0.830078125
                },
                {
                    "start": 284.32,
                    "end": 285.0,
                    "word": ".",
                    "probability": 0.9814453125
                },
                {
                    "start": 285.24,
                    "end": 285.68,
                    "word": " You",
                    "probability": 0.467529296875
                },
                {
                    "start": 285.68,
                    "end": 285.84,
                    "word": " might",
                    "probability": 0.8056640625
                },
                {
                    "start": 285.84,
                    "end": 286.0,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 286.0,
                    "end": 286.22,
                    "word": " wondering",
                    "probability": 0.77978515625
                },
                {
                    "start": 286.22,
                    "end": 286.72,
                    "word": " why",
                    "probability": 0.82080078125
                },
                {
                    "start": 286.72,
                    "end": 287.12,
                    "word": " this",
                    "probability": 0.88525390625
                },
                {
                    "start": 287.12,
                    "end": 287.46,
                    "word": " might",
                    "probability": 0.80615234375
                },
                {
                    "start": 287.46,
                    "end": 287.82,
                    "word": " be",
                    "probability": 0.912109375
                }
            ]
        },
        {
            "id": 46,
            "seek": 30660,
            "start": 287.82,
            "end": 297.12,
            "text": " damaging since the people tend to put their public information on IMDb. This could be because they",
            "tokens": [
                50756,
                25342,
                1670,
                264,
                561,
                3928,
                281,
                829,
                641,
                1908,
                1589,
                322,
                21463,
                35,
                65,
                13,
                639,
                727,
                312,
                570,
                436,
                51220
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23506943997409607,
            "compression_ratio": 1.7142857142857142,
            "no_speech_prob": 0.00045108795166015625,
            "words": [
                {
                    "start": 287.82,
                    "end": 288.28,
                    "word": " damaging",
                    "probability": 0.8779296875
                },
                {
                    "start": 288.28,
                    "end": 288.76,
                    "word": " since",
                    "probability": 0.62646484375
                },
                {
                    "start": 288.76,
                    "end": 289.28,
                    "word": " the",
                    "probability": 0.701171875
                },
                {
                    "start": 289.28,
                    "end": 291.28,
                    "word": " people",
                    "probability": 0.31884765625
                },
                {
                    "start": 291.28,
                    "end": 292.34,
                    "word": " tend",
                    "probability": 0.7763671875
                },
                {
                    "start": 292.34,
                    "end": 292.62,
                    "word": " to",
                    "probability": 0.9052734375
                },
                {
                    "start": 292.62,
                    "end": 292.96,
                    "word": " put",
                    "probability": 0.89453125
                },
                {
                    "start": 292.96,
                    "end": 293.16,
                    "word": " their",
                    "probability": 0.89306640625
                },
                {
                    "start": 293.16,
                    "end": 293.48,
                    "word": " public",
                    "probability": 0.8193359375
                },
                {
                    "start": 293.48,
                    "end": 293.92,
                    "word": " information",
                    "probability": 0.7529296875
                },
                {
                    "start": 293.92,
                    "end": 294.28,
                    "word": " on",
                    "probability": 0.896484375
                },
                {
                    "start": 294.28,
                    "end": 295.14,
                    "word": " IMDb",
                    "probability": 0.5404205322265625
                },
                {
                    "start": 295.14,
                    "end": 295.82,
                    "word": ".",
                    "probability": 0.8349609375
                },
                {
                    "start": 295.92,
                    "end": 296.4,
                    "word": " This",
                    "probability": 0.81640625
                },
                {
                    "start": 296.4,
                    "end": 296.54,
                    "word": " could",
                    "probability": 0.814453125
                },
                {
                    "start": 296.54,
                    "end": 296.68,
                    "word": " be",
                    "probability": 0.91015625
                },
                {
                    "start": 296.68,
                    "end": 296.96,
                    "word": " because",
                    "probability": 0.8408203125
                },
                {
                    "start": 296.96,
                    "end": 297.12,
                    "word": " they",
                    "probability": 0.6845703125
                }
            ]
        },
        {
            "id": 47,
            "seek": 30660,
            "start": 297.12,
            "end": 304.1,
            "text": " might be biasing or selecting a set of records to disclose on IMDb, but not on Netflix recommendation.",
            "tokens": [
                51220,
                1062,
                312,
                3228,
                3349,
                420,
                18182,
                257,
                992,
                295,
                7724,
                281,
                36146,
                322,
                21463,
                35,
                65,
                11,
                457,
                406,
                322,
                12778,
                11879,
                13,
                51564
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23506943997409607,
            "compression_ratio": 1.7142857142857142,
            "no_speech_prob": 0.00045108795166015625,
            "words": [
                {
                    "start": 297.12,
                    "end": 297.34,
                    "word": " might",
                    "probability": 0.81396484375
                },
                {
                    "start": 297.34,
                    "end": 298.0,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 298.0,
                    "end": 298.5,
                    "word": " biasing",
                    "probability": 0.8818359375
                },
                {
                    "start": 298.5,
                    "end": 298.92,
                    "word": " or",
                    "probability": 0.86962890625
                },
                {
                    "start": 298.92,
                    "end": 299.32,
                    "word": " selecting",
                    "probability": 0.5966796875
                },
                {
                    "start": 299.56,
                    "end": 299.76,
                    "word": " a set",
                    "probability": 0.89794921875
                },
                {
                    "start": 299.76,
                    "end": 300.14,
                    "word": " of",
                    "probability": 0.90576171875
                },
                {
                    "start": 300.14,
                    "end": 300.9,
                    "word": " records",
                    "probability": 0.7841796875
                },
                {
                    "start": 300.9,
                    "end": 301.16,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 301.16,
                    "end": 301.6,
                    "word": " disclose",
                    "probability": 0.79541015625
                },
                {
                    "start": 301.6,
                    "end": 301.86,
                    "word": " on",
                    "probability": 0.84375
                },
                {
                    "start": 301.86,
                    "end": 302.22,
                    "word": " IMDb",
                    "probability": 0.9459635416666666
                },
                {
                    "start": 302.22,
                    "end": 302.32,
                    "word": ",",
                    "probability": 0.441162109375
                },
                {
                    "start": 302.32,
                    "end": 302.42,
                    "word": " but",
                    "probability": 0.904296875
                },
                {
                    "start": 302.42,
                    "end": 302.6,
                    "word": " not",
                    "probability": 0.8818359375
                },
                {
                    "start": 302.6,
                    "end": 302.8,
                    "word": " on",
                    "probability": 0.88916015625
                },
                {
                    "start": 302.8,
                    "end": 303.06,
                    "word": " Netflix",
                    "probability": 0.859375
                },
                {
                    "start": 303.06,
                    "end": 303.5,
                    "word": " recommendation",
                    "probability": 0.59423828125
                },
                {
                    "start": 303.5,
                    "end": 304.1,
                    "word": ".",
                    "probability": 0.99365234375
                }
            ]
        },
        {
            "id": 48,
            "seek": 30660,
            "start": 304.1,
            "end": 309.54,
            "text": " For example, they were able to identify that for a particular person, they were able to identify",
            "tokens": [
                51564,
                1171,
                1365,
                11,
                436,
                645,
                1075,
                281,
                5876,
                300,
                337,
                257,
                1729,
                954,
                11,
                436,
                645,
                1075,
                281,
                5876,
                51844
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23506943997409607,
            "compression_ratio": 1.7142857142857142,
            "no_speech_prob": 0.00045108795166015625,
            "words": [
                {
                    "start": 304.1,
                    "end": 304.24,
                    "word": " For",
                    "probability": 0.9052734375
                },
                {
                    "start": 304.24,
                    "end": 304.52,
                    "word": " example",
                    "probability": 0.9375
                },
                {
                    "start": 304.52,
                    "end": 304.66,
                    "word": ",",
                    "probability": 0.98388671875
                },
                {
                    "start": 304.66,
                    "end": 304.78,
                    "word": " they",
                    "probability": 0.7099609375
                },
                {
                    "start": 304.78,
                    "end": 305.0,
                    "word": " were",
                    "probability": 0.8759765625
                },
                {
                    "start": 305.0,
                    "end": 305.42,
                    "word": " able",
                    "probability": 0.88671875
                },
                {
                    "start": 305.42,
                    "end": 305.64,
                    "word": " to",
                    "probability": 0.904296875
                },
                {
                    "start": 305.64,
                    "end": 306.06,
                    "word": " identify",
                    "probability": 0.896484375
                },
                {
                    "start": 306.06,
                    "end": 306.46,
                    "word": " that",
                    "probability": 0.85400390625
                },
                {
                    "start": 306.46,
                    "end": 306.94,
                    "word": " for",
                    "probability": 0.708984375
                },
                {
                    "start": 307.1,
                    "end": 307.36,
                    "word": " a particular",
                    "probability": 0.802734375
                },
                {
                    "start": 307.36,
                    "end": 308.06,
                    "word": " person",
                    "probability": 0.79248046875
                },
                {
                    "start": 308.06,
                    "end": 308.22,
                    "word": ",",
                    "probability": 0.671875
                },
                {
                    "start": 308.22,
                    "end": 308.58,
                    "word": " they",
                    "probability": 0.7314453125
                },
                {
                    "start": 308.58,
                    "end": 308.7,
                    "word": " were",
                    "probability": 0.87890625
                },
                {
                    "start": 308.7,
                    "end": 308.9,
                    "word": " able",
                    "probability": 0.88623046875
                },
                {
                    "start": 308.9,
                    "end": 309.14,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 309.14,
                    "end": 309.54,
                    "word": " identify",
                    "probability": 0.896484375
                }
            ]
        },
        {
            "id": 49,
            "seek": 33520,
            "start": 309.54,
            "end": 315.96,
            "text": " just based on his Netflix record, that he might have a voting preference from his voting record on",
            "tokens": [
                50364,
                445,
                2361,
                322,
                702,
                12778,
                2136,
                11,
                300,
                415,
                1062,
                362,
                257,
                10419,
                17502,
                490,
                702,
                10419,
                2136,
                322,
                50680
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2956730697908972,
            "compression_ratio": 1.6866666666666668,
            "no_speech_prob": 0.0006165504455566406,
            "words": [
                {
                    "start": 309.54,
                    "end": 309.8,
                    "word": " just",
                    "probability": 0.1524658203125
                },
                {
                    "start": 309.8,
                    "end": 310.1,
                    "word": " based",
                    "probability": 0.79931640625
                },
                {
                    "start": 310.1,
                    "end": 310.22,
                    "word": " on",
                    "probability": 0.89208984375
                },
                {
                    "start": 310.22,
                    "end": 310.34,
                    "word": " his",
                    "probability": 0.7177734375
                },
                {
                    "start": 310.34,
                    "end": 310.58,
                    "word": " Netflix",
                    "probability": 0.8349609375
                },
                {
                    "start": 310.58,
                    "end": 311.04,
                    "word": " record",
                    "probability": 0.81982421875
                },
                {
                    "start": 311.04,
                    "end": 311.22,
                    "word": ",",
                    "probability": 0.305419921875
                },
                {
                    "start": 311.22,
                    "end": 311.5,
                    "word": " that",
                    "probability": 0.82177734375
                },
                {
                    "start": 311.5,
                    "end": 311.88,
                    "word": " he",
                    "probability": 0.8251953125
                },
                {
                    "start": 311.88,
                    "end": 312.12,
                    "word": " might",
                    "probability": 0.4248046875
                },
                {
                    "start": 312.12,
                    "end": 312.3,
                    "word": " have",
                    "probability": 0.87548828125
                },
                {
                    "start": 312.4,
                    "end": 312.56,
                    "word": " a voting",
                    "probability": 0.681640625
                },
                {
                    "start": 312.56,
                    "end": 313.16,
                    "word": " preference",
                    "probability": 0.81298828125
                },
                {
                    "start": 313.16,
                    "end": 313.64,
                    "word": " from",
                    "probability": 0.74609375
                },
                {
                    "start": 313.64,
                    "end": 314.06,
                    "word": " his",
                    "probability": 0.900390625
                },
                {
                    "start": 314.06,
                    "end": 315.24,
                    "word": " voting",
                    "probability": 0.5712890625
                },
                {
                    "start": 315.24,
                    "end": 315.6,
                    "word": " record",
                    "probability": 0.904296875
                },
                {
                    "start": 315.6,
                    "end": 315.96,
                    "word": " on",
                    "probability": 0.8935546875
                }
            ]
        },
        {
            "id": 50,
            "seek": 33520,
            "start": 315.96,
            "end": 321.02,
            "text": " Power and Terror, Norm Chomsky and Our Times, and Periodic 9-11. They might be able to infer on his political",
            "tokens": [
                50680,
                7086,
                293,
                36174,
                11,
                8702,
                761,
                4785,
                4133,
                293,
                2621,
                11366,
                11,
                293,
                34976,
                299,
                1722,
                12,
                5348,
                13,
                814,
                1062,
                312,
                1075,
                281,
                13596,
                322,
                702,
                3905,
                50944
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2956730697908972,
            "compression_ratio": 1.6866666666666668,
            "no_speech_prob": 0.0006165504455566406,
            "words": [
                {
                    "start": 315.96,
                    "end": 316.24,
                    "word": " Power",
                    "probability": 0.476318359375
                },
                {
                    "start": 316.24,
                    "end": 316.42,
                    "word": " and",
                    "probability": 0.7333984375
                },
                {
                    "start": 316.42,
                    "end": 316.64,
                    "word": " Terror",
                    "probability": 0.9189453125
                },
                {
                    "start": 316.64,
                    "end": 316.74,
                    "word": ",",
                    "probability": 0.7197265625
                },
                {
                    "start": 316.74,
                    "end": 316.92,
                    "word": " Norm",
                    "probability": 0.50390625
                },
                {
                    "start": 316.92,
                    "end": 317.28,
                    "word": " Chomsky",
                    "probability": 0.9202473958333334
                },
                {
                    "start": 317.28,
                    "end": 317.38,
                    "word": " and",
                    "probability": 0.448486328125
                },
                {
                    "start": 317.38,
                    "end": 317.48,
                    "word": " Our",
                    "probability": 0.1566162109375
                },
                {
                    "start": 317.48,
                    "end": 317.76,
                    "word": " Times",
                    "probability": 0.6728515625
                },
                {
                    "start": 317.76,
                    "end": 317.8,
                    "word": ",",
                    "probability": 0.36279296875
                },
                {
                    "start": 317.8,
                    "end": 317.9,
                    "word": " and",
                    "probability": 0.258544921875
                },
                {
                    "start": 317.9,
                    "end": 318.18,
                    "word": " Periodic",
                    "probability": 0.5181884765625
                },
                {
                    "start": 318.18,
                    "end": 318.36,
                    "word": " 9",
                    "probability": 0.61962890625
                },
                {
                    "start": 318.36,
                    "end": 318.62,
                    "word": "-11",
                    "probability": 0.540740966796875
                },
                {
                    "start": 318.62,
                    "end": 319.14,
                    "word": ".",
                    "probability": 0.81982421875
                },
                {
                    "start": 319.14,
                    "end": 319.32,
                    "word": " They",
                    "probability": 0.67626953125
                },
                {
                    "start": 319.32,
                    "end": 319.48,
                    "word": " might",
                    "probability": 0.81689453125
                },
                {
                    "start": 319.48,
                    "end": 319.6,
                    "word": " be",
                    "probability": 0.90576171875
                },
                {
                    "start": 319.6,
                    "end": 319.74,
                    "word": " able",
                    "probability": 0.88623046875
                },
                {
                    "start": 319.74,
                    "end": 319.92,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 319.92,
                    "end": 320.26,
                    "word": " infer",
                    "probability": 0.84033203125
                },
                {
                    "start": 320.26,
                    "end": 320.46,
                    "word": " on",
                    "probability": 0.88525390625
                },
                {
                    "start": 320.46,
                    "end": 320.64,
                    "word": " his",
                    "probability": 0.89501953125
                },
                {
                    "start": 320.64,
                    "end": 321.02,
                    "word": " political",
                    "probability": 0.78564453125
                }
            ]
        },
        {
            "id": 51,
            "seek": 33520,
            "start": 321.02,
            "end": 326.08,
            "text": " preference based on the Jesus of Nazareth and Gospel of John ratings, and also his eating habit based",
            "tokens": [
                50944,
                17502,
                2361,
                322,
                264,
                2705,
                295,
                11870,
                40414,
                293,
                23163,
                295,
                2619,
                24603,
                11,
                293,
                611,
                702,
                3936,
                7164,
                2361,
                51192
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2956730697908972,
            "compression_ratio": 1.6866666666666668,
            "no_speech_prob": 0.0006165504455566406,
            "words": [
                {
                    "start": 321.02,
                    "end": 321.6,
                    "word": " preference",
                    "probability": 0.7431640625
                },
                {
                    "start": 321.6,
                    "end": 321.86,
                    "word": " based",
                    "probability": 0.705078125
                },
                {
                    "start": 321.86,
                    "end": 321.98,
                    "word": " on",
                    "probability": 0.90234375
                },
                {
                    "start": 321.98,
                    "end": 322.08,
                    "word": " the",
                    "probability": 0.609375
                },
                {
                    "start": 322.08,
                    "end": 322.28,
                    "word": " Jesus",
                    "probability": 0.8671875
                },
                {
                    "start": 322.28,
                    "end": 322.54,
                    "word": " of",
                    "probability": 0.87890625
                },
                {
                    "start": 322.54,
                    "end": 323.0,
                    "word": " Nazareth",
                    "probability": 0.88134765625
                },
                {
                    "start": 323.0,
                    "end": 323.18,
                    "word": " and",
                    "probability": 0.83349609375
                },
                {
                    "start": 323.18,
                    "end": 323.44,
                    "word": " Gospel",
                    "probability": 0.6201171875
                },
                {
                    "start": 323.44,
                    "end": 323.6,
                    "word": " of",
                    "probability": 0.88720703125
                },
                {
                    "start": 323.6,
                    "end": 323.74,
                    "word": " John",
                    "probability": 0.79150390625
                },
                {
                    "start": 323.74,
                    "end": 324.12,
                    "word": " ratings",
                    "probability": 0.5546875
                },
                {
                    "start": 324.12,
                    "end": 324.42,
                    "word": ",",
                    "probability": 0.57861328125
                },
                {
                    "start": 324.42,
                    "end": 324.64,
                    "word": " and",
                    "probability": 0.8935546875
                },
                {
                    "start": 324.64,
                    "end": 324.88,
                    "word": " also",
                    "probability": 0.81494140625
                },
                {
                    "start": 324.88,
                    "end": 325.14,
                    "word": " his",
                    "probability": 0.853515625
                },
                {
                    "start": 325.14,
                    "end": 325.38,
                    "word": " eating",
                    "probability": 0.86181640625
                },
                {
                    "start": 325.38,
                    "end": 325.72,
                    "word": " habit",
                    "probability": 0.68603515625
                },
                {
                    "start": 325.72,
                    "end": 326.08,
                    "word": " based",
                    "probability": 0.7734375
                }
            ]
        },
        {
            "id": 52,
            "seek": 33520,
            "start": 326.08,
            "end": 333.32,
            "text": " on his recommendations on supersizing. In the next slide, we look at a new type of attack called",
            "tokens": [
                51192,
                322,
                702,
                10434,
                322,
                37906,
                3319,
                13,
                682,
                264,
                958,
                4137,
                11,
                321,
                574,
                412,
                257,
                777,
                2010,
                295,
                2690,
                1219,
                51548
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2956730697908972,
            "compression_ratio": 1.6866666666666668,
            "no_speech_prob": 0.0006165504455566406,
            "words": [
                {
                    "start": 326.08,
                    "end": 326.42,
                    "word": " on",
                    "probability": 0.90625
                },
                {
                    "start": 326.42,
                    "end": 326.9,
                    "word": " his",
                    "probability": 0.8984375
                },
                {
                    "start": 326.9,
                    "end": 327.62,
                    "word": " recommendations",
                    "probability": 0.6474609375
                },
                {
                    "start": 327.62,
                    "end": 328.06,
                    "word": " on",
                    "probability": 0.88134765625
                },
                {
                    "start": 328.06,
                    "end": 328.64,
                    "word": " supersizing",
                    "probability": 0.6522216796875
                },
                {
                    "start": 328.64,
                    "end": 329.16,
                    "word": ".",
                    "probability": 0.97900390625
                },
                {
                    "start": 329.58,
                    "end": 330.1,
                    "word": " In",
                    "probability": 0.84375
                },
                {
                    "start": 330.1,
                    "end": 330.22,
                    "word": " the",
                    "probability": 0.810546875
                },
                {
                    "start": 330.22,
                    "end": 330.5,
                    "word": " next",
                    "probability": 0.8603515625
                },
                {
                    "start": 330.5,
                    "end": 331.36,
                    "word": " slide",
                    "probability": 0.87890625
                },
                {
                    "start": 331.36,
                    "end": 331.46,
                    "word": ",",
                    "probability": 0.77734375
                },
                {
                    "start": 331.46,
                    "end": 331.6,
                    "word": " we",
                    "probability": 0.8974609375
                },
                {
                    "start": 331.6,
                    "end": 331.86,
                    "word": " look",
                    "probability": 0.90869140625
                },
                {
                    "start": 331.86,
                    "end": 332.08,
                    "word": " at",
                    "probability": 0.90625
                },
                {
                    "start": 332.26,
                    "end": 332.46,
                    "word": " a new",
                    "probability": 0.8271484375
                },
                {
                    "start": 332.46,
                    "end": 332.78,
                    "word": " type",
                    "probability": 0.916015625
                },
                {
                    "start": 332.78,
                    "end": 332.94,
                    "word": " of",
                    "probability": 0.91064453125
                },
                {
                    "start": 332.94,
                    "end": 333.1,
                    "word": " attack",
                    "probability": 0.73046875
                },
                {
                    "start": 333.1,
                    "end": 333.32,
                    "word": " called",
                    "probability": 0.74755859375
                }
            ]
        },
        {
            "id": 53,
            "seek": 33520,
            "start": 333.32,
            "end": 338.14,
            "text": " the reconstruction attack. This is more prone to models that tend to keep the training information",
            "tokens": [
                51548,
                264,
                31565,
                2690,
                13,
                639,
                307,
                544,
                25806,
                281,
                5245,
                300,
                3928,
                281,
                1066,
                264,
                3097,
                1589,
                51800
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2956730697908972,
            "compression_ratio": 1.6866666666666668,
            "no_speech_prob": 0.0006165504455566406,
            "words": [
                {
                    "start": 333.32,
                    "end": 333.42,
                    "word": " the",
                    "probability": 0.351806640625
                },
                {
                    "start": 333.42,
                    "end": 333.78,
                    "word": " reconstruction",
                    "probability": 0.76416015625
                },
                {
                    "start": 333.78,
                    "end": 334.26,
                    "word": " attack",
                    "probability": 0.81005859375
                },
                {
                    "start": 334.26,
                    "end": 334.68,
                    "word": ".",
                    "probability": 0.97509765625
                },
                {
                    "start": 334.68,
                    "end": 334.88,
                    "word": " This",
                    "probability": 0.8232421875
                },
                {
                    "start": 334.88,
                    "end": 335.02,
                    "word": " is",
                    "probability": 0.9091796875
                },
                {
                    "start": 335.02,
                    "end": 335.2,
                    "word": " more",
                    "probability": 0.88720703125
                },
                {
                    "start": 335.2,
                    "end": 335.46,
                    "word": " prone",
                    "probability": 0.498291015625
                },
                {
                    "start": 335.46,
                    "end": 335.66,
                    "word": " to",
                    "probability": 0.91064453125
                },
                {
                    "start": 335.66,
                    "end": 335.92,
                    "word": " models",
                    "probability": 0.58056640625
                },
                {
                    "start": 335.92,
                    "end": 336.16,
                    "word": " that",
                    "probability": 0.8798828125
                },
                {
                    "start": 336.16,
                    "end": 336.42,
                    "word": " tend",
                    "probability": 0.85205078125
                },
                {
                    "start": 336.42,
                    "end": 336.6,
                    "word": " to",
                    "probability": 0.9091796875
                },
                {
                    "start": 336.6,
                    "end": 336.92,
                    "word": " keep",
                    "probability": 0.80419921875
                },
                {
                    "start": 336.92,
                    "end": 337.44,
                    "word": " the",
                    "probability": 0.80908203125
                },
                {
                    "start": 337.44,
                    "end": 337.74,
                    "word": " training",
                    "probability": 0.666015625
                },
                {
                    "start": 337.74,
                    "end": 338.14,
                    "word": " information",
                    "probability": 0.74365234375
                }
            ]
        },
        {
            "id": 54,
            "seek": 36240,
            "start": 338.14,
            "end": 345.56,
            "text": " in the structure itself, such as near-neighbor classifier and kernel-based SVM. But this is also",
            "tokens": [
                50368,
                294,
                264,
                3877,
                2564,
                11,
                1270,
                382,
                2651,
                12,
                716,
                910,
                3918,
                1508,
                9902,
                293,
                28256,
                12,
                6032,
                31910,
                44,
                13,
                583,
                341,
                307,
                611,
                50736
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2814295956458168,
            "compression_ratio": 1.6085106382978724,
            "no_speech_prob": 0.0002868175506591797,
            "words": [
                {
                    "start": 338.14,
                    "end": 338.28,
                    "word": " in",
                    "probability": 0.59375
                },
                {
                    "start": 338.28,
                    "end": 339.14,
                    "word": " the",
                    "probability": 0.8056640625
                },
                {
                    "start": 339.14,
                    "end": 339.62,
                    "word": " structure",
                    "probability": 0.74560546875
                },
                {
                    "start": 339.62,
                    "end": 340.02,
                    "word": " itself",
                    "probability": 0.791015625
                },
                {
                    "start": 340.02,
                    "end": 340.1,
                    "word": ",",
                    "probability": 0.595703125
                },
                {
                    "start": 340.1,
                    "end": 340.28,
                    "word": " such",
                    "probability": 0.87060546875
                },
                {
                    "start": 340.28,
                    "end": 340.5,
                    "word": " as",
                    "probability": 0.91259765625
                },
                {
                    "start": 340.5,
                    "end": 340.72,
                    "word": " near",
                    "probability": 0.310302734375
                },
                {
                    "start": 340.72,
                    "end": 341.38,
                    "word": "-neighbor",
                    "probability": 0.757568359375
                },
                {
                    "start": 341.38,
                    "end": 342.02,
                    "word": " classifier",
                    "probability": 0.916748046875
                },
                {
                    "start": 342.02,
                    "end": 342.9,
                    "word": " and",
                    "probability": 0.73095703125
                },
                {
                    "start": 342.9,
                    "end": 343.14,
                    "word": " kernel",
                    "probability": 0.77099609375
                },
                {
                    "start": 343.14,
                    "end": 343.44,
                    "word": "-based",
                    "probability": 0.8408203125
                },
                {
                    "start": 343.44,
                    "end": 344.1,
                    "word": " SVM",
                    "probability": 0.839111328125
                },
                {
                    "start": 344.1,
                    "end": 344.7,
                    "word": ".",
                    "probability": 0.91015625
                },
                {
                    "start": 344.76,
                    "end": 344.94,
                    "word": " But",
                    "probability": 0.86669921875
                },
                {
                    "start": 344.94,
                    "end": 345.16,
                    "word": " this",
                    "probability": 0.873046875
                },
                {
                    "start": 345.16,
                    "end": 345.32,
                    "word": " is",
                    "probability": 0.90625
                },
                {
                    "start": 345.32,
                    "end": 345.56,
                    "word": " also",
                    "probability": 0.822265625
                }
            ]
        },
        {
            "id": 55,
            "seek": 36240,
            "start": 345.56,
            "end": 351.76,
            "text": " prone to neural networks as well, as we can see in the next slide, where in production environments,",
            "tokens": [
                50736,
                25806,
                281,
                18161,
                9590,
                382,
                731,
                11,
                382,
                321,
                393,
                536,
                294,
                264,
                958,
                4137,
                11,
                689,
                294,
                4265,
                12388,
                11,
                51000
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2814295956458168,
            "compression_ratio": 1.6085106382978724,
            "no_speech_prob": 0.0002868175506591797,
            "words": [
                {
                    "start": 345.56,
                    "end": 345.96,
                    "word": " prone",
                    "probability": 0.833984375
                },
                {
                    "start": 345.96,
                    "end": 346.44,
                    "word": " to",
                    "probability": 0.9033203125
                },
                {
                    "start": 346.44,
                    "end": 346.72,
                    "word": " neural",
                    "probability": 0.4453125
                },
                {
                    "start": 346.72,
                    "end": 346.98,
                    "word": " networks",
                    "probability": 0.5068359375
                },
                {
                    "start": 346.98,
                    "end": 347.18,
                    "word": " as",
                    "probability": 0.84130859375
                },
                {
                    "start": 347.18,
                    "end": 347.38,
                    "word": " well",
                    "probability": 0.87451171875
                },
                {
                    "start": 347.38,
                    "end": 347.48,
                    "word": ",",
                    "probability": 0.6611328125
                },
                {
                    "start": 347.48,
                    "end": 347.6,
                    "word": " as",
                    "probability": 0.9052734375
                },
                {
                    "start": 347.6,
                    "end": 347.72,
                    "word": " we",
                    "probability": 0.89453125
                },
                {
                    "start": 347.72,
                    "end": 347.82,
                    "word": " can",
                    "probability": 0.374267578125
                },
                {
                    "start": 347.82,
                    "end": 347.98,
                    "word": " see",
                    "probability": 0.8193359375
                },
                {
                    "start": 347.98,
                    "end": 348.06,
                    "word": " in",
                    "probability": 0.8828125
                },
                {
                    "start": 348.06,
                    "end": 348.16,
                    "word": " the",
                    "probability": 0.8115234375
                },
                {
                    "start": 348.16,
                    "end": 348.34,
                    "word": " next",
                    "probability": 0.845703125
                },
                {
                    "start": 348.34,
                    "end": 348.54,
                    "word": " slide",
                    "probability": 0.8994140625
                },
                {
                    "start": 348.54,
                    "end": 348.66,
                    "word": ",",
                    "probability": 0.4990234375
                },
                {
                    "start": 348.66,
                    "end": 349.02,
                    "word": " where",
                    "probability": 0.89306640625
                },
                {
                    "start": 349.02,
                    "end": 349.94,
                    "word": " in",
                    "probability": 0.5185546875
                },
                {
                    "start": 349.94,
                    "end": 350.24,
                    "word": " production",
                    "probability": 0.82568359375
                },
                {
                    "start": 350.24,
                    "end": 350.9,
                    "word": " environments",
                    "probability": 0.66552734375
                },
                {
                    "start": 350.9,
                    "end": 351.76,
                    "word": ",",
                    "probability": 0.8720703125
                }
            ]
        },
        {
            "id": 56,
            "seek": 36240,
            "start": 351.76,
            "end": 357.82,
            "text": " researchers and even implementers tend to reuse neural network models just because of something",
            "tokens": [
                51036,
                10309,
                293,
                754,
                4445,
                433,
                3928,
                281,
                26225,
                18161,
                3209,
                5245,
                445,
                570,
                295,
                746,
                51348
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2814295956458168,
            "compression_ratio": 1.6085106382978724,
            "no_speech_prob": 0.0002868175506591797,
            "words": [
                {
                    "start": 351.76,
                    "end": 352.12,
                    "word": " researchers",
                    "probability": 0.78515625
                },
                {
                    "start": 352.12,
                    "end": 352.5,
                    "word": " and",
                    "probability": 0.7958984375
                },
                {
                    "start": 352.5,
                    "end": 352.86,
                    "word": " even",
                    "probability": 0.81396484375
                },
                {
                    "start": 352.86,
                    "end": 353.48,
                    "word": " implementers",
                    "probability": 0.81103515625
                },
                {
                    "start": 353.48,
                    "end": 353.72,
                    "word": " tend",
                    "probability": 0.73095703125
                },
                {
                    "start": 353.72,
                    "end": 353.9,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 353.9,
                    "end": 354.36,
                    "word": " reuse",
                    "probability": 0.75048828125
                },
                {
                    "start": 354.36,
                    "end": 355.68,
                    "word": " neural",
                    "probability": 0.84765625
                },
                {
                    "start": 355.68,
                    "end": 355.92,
                    "word": " network",
                    "probability": 0.68212890625
                },
                {
                    "start": 355.92,
                    "end": 356.2,
                    "word": " models",
                    "probability": 0.79296875
                },
                {
                    "start": 356.2,
                    "end": 356.44,
                    "word": " just",
                    "probability": 0.71533203125
                },
                {
                    "start": 356.44,
                    "end": 356.8,
                    "word": " because",
                    "probability": 0.84619140625
                },
                {
                    "start": 356.8,
                    "end": 357.4,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 357.4,
                    "end": 357.82,
                    "word": " something",
                    "probability": 0.78515625
                }
            ]
        },
        {
            "id": 57,
            "seek": 36240,
            "start": 357.82,
            "end": 365.34,
            "text": " called style transfer, or style transfer where they're able to utilize one model and",
            "tokens": [
                51348,
                1219,
                3758,
                5003,
                11,
                420,
                3758,
                5003,
                689,
                436,
                434,
                1075,
                281,
                16117,
                472,
                2316,
                293,
                51724
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2814295956458168,
            "compression_ratio": 1.6085106382978724,
            "no_speech_prob": 0.0002868175506591797,
            "words": [
                {
                    "start": 357.82,
                    "end": 358.3,
                    "word": " called",
                    "probability": 0.81005859375
                },
                {
                    "start": 358.3,
                    "end": 359.24,
                    "word": " style",
                    "probability": 0.68798828125
                },
                {
                    "start": 359.24,
                    "end": 359.74,
                    "word": " transfer",
                    "probability": 0.79541015625
                },
                {
                    "start": 359.74,
                    "end": 360.58,
                    "word": ",",
                    "probability": 0.365478515625
                },
                {
                    "start": 360.58,
                    "end": 360.58,
                    "word": " or",
                    "probability": 0.6982421875
                },
                {
                    "start": 360.58,
                    "end": 362.06,
                    "word": " style",
                    "probability": 0.40771484375
                },
                {
                    "start": 362.06,
                    "end": 362.46,
                    "word": " transfer",
                    "probability": 0.82763671875
                },
                {
                    "start": 362.46,
                    "end": 362.68,
                    "word": " where",
                    "probability": 0.7734375
                },
                {
                    "start": 362.68,
                    "end": 362.84,
                    "word": " they're",
                    "probability": 0.79296875
                },
                {
                    "start": 362.84,
                    "end": 363.0,
                    "word": " able",
                    "probability": 0.888671875
                },
                {
                    "start": 363.0,
                    "end": 363.4,
                    "word": " to",
                    "probability": 0.90283203125
                },
                {
                    "start": 363.4,
                    "end": 364.06,
                    "word": " utilize",
                    "probability": 0.69384765625
                },
                {
                    "start": 364.06,
                    "end": 364.7,
                    "word": " one",
                    "probability": 0.80859375
                },
                {
                    "start": 364.7,
                    "end": 365.0,
                    "word": " model",
                    "probability": 0.79736328125
                },
                {
                    "start": 365.0,
                    "end": 365.34,
                    "word": " and",
                    "probability": 0.87744140625
                }
            ]
        },
        {
            "id": 58,
            "seek": 39118,
            "start": 367.3,
            "end": 371.9,
            "text": " create a new prediction from that same model using the inputs and architecture of it,",
            "tokens": [
                50372,
                1884,
                257,
                777,
                17630,
                490,
                300,
                912,
                2316,
                1228,
                264,
                15743,
                293,
                9482,
                295,
                309,
                11,
                50592
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3034620051290475,
            "compression_ratio": 1.7683397683397684,
            "no_speech_prob": 8.219480514526367e-05,
            "words": [
                {
                    "start": 367.3,
                    "end": 367.86,
                    "word": " create",
                    "probability": 0.0225372314453125
                },
                {
                    "start": 368.14,
                    "end": 368.36,
                    "word": " a new",
                    "probability": 0.8359375
                },
                {
                    "start": 368.36,
                    "end": 368.7,
                    "word": " prediction",
                    "probability": 0.841796875
                },
                {
                    "start": 368.7,
                    "end": 369.0,
                    "word": " from",
                    "probability": 0.7451171875
                },
                {
                    "start": 369.0,
                    "end": 369.2,
                    "word": " that",
                    "probability": 0.87353515625
                },
                {
                    "start": 369.2,
                    "end": 369.42,
                    "word": " same",
                    "probability": 0.79638671875
                },
                {
                    "start": 369.42,
                    "end": 369.7,
                    "word": " model",
                    "probability": 0.78662109375
                },
                {
                    "start": 369.7,
                    "end": 370.02,
                    "word": " using",
                    "probability": 0.47900390625
                },
                {
                    "start": 370.02,
                    "end": 370.18,
                    "word": " the",
                    "probability": 0.80517578125
                },
                {
                    "start": 370.18,
                    "end": 370.54,
                    "word": " inputs",
                    "probability": 0.71044921875
                },
                {
                    "start": 370.54,
                    "end": 370.76,
                    "word": " and",
                    "probability": 0.80908203125
                },
                {
                    "start": 370.76,
                    "end": 371.22,
                    "word": " architecture",
                    "probability": 0.7451171875
                },
                {
                    "start": 371.22,
                    "end": 371.42,
                    "word": " of",
                    "probability": 0.89208984375
                },
                {
                    "start": 371.42,
                    "end": 371.6,
                    "word": " it",
                    "probability": 0.9091796875
                },
                {
                    "start": 371.6,
                    "end": 371.9,
                    "word": ",",
                    "probability": 0.261962890625
                }
            ]
        },
        {
            "id": 59,
            "seek": 39118,
            "start": 371.9,
            "end": 377.86,
            "text": " and reconstruct back to the original input. In the next slide, we also see a similar type of",
            "tokens": [
                50592,
                293,
                31499,
                646,
                281,
                264,
                3380,
                4846,
                13,
                682,
                264,
                958,
                4137,
                11,
                321,
                611,
                536,
                257,
                2531,
                2010,
                295,
                50884
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3034620051290475,
            "compression_ratio": 1.7683397683397684,
            "no_speech_prob": 8.219480514526367e-05,
            "words": [
                {
                    "start": 371.9,
                    "end": 371.96,
                    "word": " and",
                    "probability": 0.8515625
                },
                {
                    "start": 371.96,
                    "end": 372.2,
                    "word": " reconstruct",
                    "probability": 0.60009765625
                },
                {
                    "start": 372.2,
                    "end": 372.58,
                    "word": " back",
                    "probability": 0.78271484375
                },
                {
                    "start": 372.58,
                    "end": 372.74,
                    "word": " to",
                    "probability": 0.908203125
                },
                {
                    "start": 372.74,
                    "end": 372.86,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 372.86,
                    "end": 373.1,
                    "word": " original",
                    "probability": 0.8330078125
                },
                {
                    "start": 373.1,
                    "end": 373.44,
                    "word": " input",
                    "probability": 0.7724609375
                },
                {
                    "start": 373.44,
                    "end": 374.0,
                    "word": ".",
                    "probability": 0.97900390625
                },
                {
                    "start": 374.5,
                    "end": 375.0,
                    "word": " In",
                    "probability": 0.88427734375
                },
                {
                    "start": 375.0,
                    "end": 375.12,
                    "word": " the",
                    "probability": 0.8125
                },
                {
                    "start": 375.12,
                    "end": 375.32,
                    "word": " next",
                    "probability": 0.85693359375
                },
                {
                    "start": 375.32,
                    "end": 375.62,
                    "word": " slide",
                    "probability": 0.88818359375
                },
                {
                    "start": 375.62,
                    "end": 375.74,
                    "word": ",",
                    "probability": 0.77978515625
                },
                {
                    "start": 375.74,
                    "end": 375.84,
                    "word": " we",
                    "probability": 0.85888671875
                },
                {
                    "start": 375.84,
                    "end": 376.16,
                    "word": " also",
                    "probability": 0.63671875
                },
                {
                    "start": 376.16,
                    "end": 376.44,
                    "word": " see",
                    "probability": 0.8203125
                },
                {
                    "start": 376.72,
                    "end": 377.42,
                    "word": " a similar",
                    "probability": 0.9228515625
                },
                {
                    "start": 377.42,
                    "end": 377.64,
                    "word": " type",
                    "probability": 0.927734375
                },
                {
                    "start": 377.64,
                    "end": 377.86,
                    "word": " of",
                    "probability": 0.90673828125
                }
            ]
        },
        {
            "id": 60,
            "seek": 39118,
            "start": 377.86,
            "end": 383.7,
            "text": " attack called model inversion attack, where in this case, a feature such as an image is fed into",
            "tokens": [
                50884,
                2690,
                1219,
                2316,
                43576,
                2690,
                11,
                689,
                294,
                341,
                1389,
                11,
                257,
                4111,
                1270,
                382,
                364,
                3256,
                307,
                4636,
                666,
                51168
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3034620051290475,
            "compression_ratio": 1.7683397683397684,
            "no_speech_prob": 8.219480514526367e-05,
            "words": [
                {
                    "start": 377.86,
                    "end": 378.28,
                    "word": " attack",
                    "probability": 0.544921875
                },
                {
                    "start": 378.28,
                    "end": 378.62,
                    "word": " called",
                    "probability": 0.73681640625
                },
                {
                    "start": 378.62,
                    "end": 378.86,
                    "word": " model",
                    "probability": 0.7578125
                },
                {
                    "start": 378.86,
                    "end": 379.22,
                    "word": " inversion",
                    "probability": 0.54345703125
                },
                {
                    "start": 379.22,
                    "end": 379.62,
                    "word": " attack",
                    "probability": 0.8369140625
                },
                {
                    "start": 379.62,
                    "end": 379.92,
                    "word": ",",
                    "probability": 0.66162109375
                },
                {
                    "start": 379.92,
                    "end": 380.26,
                    "word": " where",
                    "probability": 0.89599609375
                },
                {
                    "start": 380.26,
                    "end": 380.74,
                    "word": " in",
                    "probability": 0.68603515625
                },
                {
                    "start": 380.74,
                    "end": 380.94,
                    "word": " this",
                    "probability": 0.88232421875
                },
                {
                    "start": 380.94,
                    "end": 381.38,
                    "word": " case",
                    "probability": 0.82568359375
                },
                {
                    "start": 381.38,
                    "end": 381.82,
                    "word": ",",
                    "probability": 0.79150390625
                },
                {
                    "start": 382.02,
                    "end": 382.3,
                    "word": " a feature",
                    "probability": 0.8486328125
                },
                {
                    "start": 382.3,
                    "end": 382.52,
                    "word": " such",
                    "probability": 0.83935546875
                },
                {
                    "start": 382.52,
                    "end": 382.66,
                    "word": " as",
                    "probability": 0.90234375
                },
                {
                    "start": 382.66,
                    "end": 382.82,
                    "word": " an",
                    "probability": 0.88916015625
                },
                {
                    "start": 382.82,
                    "end": 383.12,
                    "word": " image",
                    "probability": 0.84423828125
                },
                {
                    "start": 383.12,
                    "end": 383.28,
                    "word": " is",
                    "probability": 0.8271484375
                },
                {
                    "start": 383.28,
                    "end": 383.44,
                    "word": " fed",
                    "probability": 0.720703125
                },
                {
                    "start": 383.44,
                    "end": 383.7,
                    "word": " into",
                    "probability": 0.689453125
                }
            ]
        },
        {
            "id": 61,
            "seek": 39118,
            "start": 383.86,
            "end": 388.82,
            "text": " a neural network model and produces a set of probabilities, and then the researcher can use",
            "tokens": [
                51168,
                257,
                18161,
                3209,
                2316,
                293,
                14725,
                257,
                992,
                295,
                33783,
                11,
                293,
                550,
                264,
                21751,
                393,
                764,
                51416
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3034620051290475,
            "compression_ratio": 1.7683397683397684,
            "no_speech_prob": 8.219480514526367e-05,
            "words": [
                {
                    "start": 383.86,
                    "end": 384.0,
                    "word": " a neural",
                    "probability": 0.92529296875
                },
                {
                    "start": 384.0,
                    "end": 384.3,
                    "word": " network",
                    "probability": 0.841796875
                },
                {
                    "start": 384.3,
                    "end": 384.7,
                    "word": " model",
                    "probability": 0.81396484375
                },
                {
                    "start": 384.7,
                    "end": 385.32,
                    "word": " and",
                    "probability": 0.31689453125
                },
                {
                    "start": 385.32,
                    "end": 385.64,
                    "word": " produces",
                    "probability": 0.72509765625
                },
                {
                    "start": 385.88,
                    "end": 385.98,
                    "word": " a set",
                    "probability": 0.90234375
                },
                {
                    "start": 385.98,
                    "end": 386.12,
                    "word": " of",
                    "probability": 0.90087890625
                },
                {
                    "start": 386.12,
                    "end": 386.58,
                    "word": " probabilities",
                    "probability": 0.8203125
                },
                {
                    "start": 386.58,
                    "end": 386.94,
                    "word": ",",
                    "probability": 0.2900390625
                },
                {
                    "start": 386.94,
                    "end": 387.02,
                    "word": " and",
                    "probability": 0.88818359375
                },
                {
                    "start": 387.02,
                    "end": 387.3,
                    "word": " then",
                    "probability": 0.7294921875
                },
                {
                    "start": 387.3,
                    "end": 387.78,
                    "word": " the",
                    "probability": 0.436279296875
                },
                {
                    "start": 387.78,
                    "end": 388.06,
                    "word": " researcher",
                    "probability": 0.8173828125
                },
                {
                    "start": 388.06,
                    "end": 388.64,
                    "word": " can",
                    "probability": 0.90087890625
                },
                {
                    "start": 388.64,
                    "end": 388.82,
                    "word": " use",
                    "probability": 0.8203125
                }
            ]
        },
        {
            "id": 62,
            "seek": 39118,
            "start": 388.82,
            "end": 394.12,
            "text": " these probabilities and name and reconstruct an image back to the real thing. For example,",
            "tokens": [
                51416,
                613,
                33783,
                293,
                1315,
                293,
                31499,
                364,
                3256,
                646,
                281,
                264,
                957,
                551,
                13,
                1171,
                1365,
                11,
                51692
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3034620051290475,
            "compression_ratio": 1.7683397683397684,
            "no_speech_prob": 8.219480514526367e-05,
            "words": [
                {
                    "start": 388.82,
                    "end": 389.02,
                    "word": " these",
                    "probability": 0.69873046875
                },
                {
                    "start": 389.02,
                    "end": 389.48,
                    "word": " probabilities",
                    "probability": 0.8525390625
                },
                {
                    "start": 389.48,
                    "end": 389.9,
                    "word": " and",
                    "probability": 0.81982421875
                },
                {
                    "start": 389.9,
                    "end": 390.14,
                    "word": " name",
                    "probability": 0.424560546875
                },
                {
                    "start": 390.14,
                    "end": 390.38,
                    "word": " and",
                    "probability": 0.75732421875
                },
                {
                    "start": 390.38,
                    "end": 390.64,
                    "word": " reconstruct",
                    "probability": 0.8349609375
                },
                {
                    "start": 390.64,
                    "end": 391.1,
                    "word": " an",
                    "probability": 0.426513671875
                },
                {
                    "start": 391.1,
                    "end": 391.28,
                    "word": " image",
                    "probability": 0.8359375
                },
                {
                    "start": 391.28,
                    "end": 391.62,
                    "word": " back",
                    "probability": 0.8037109375
                },
                {
                    "start": 391.62,
                    "end": 391.86,
                    "word": " to",
                    "probability": 0.8974609375
                },
                {
                    "start": 391.86,
                    "end": 391.98,
                    "word": " the",
                    "probability": 0.8154296875
                },
                {
                    "start": 391.98,
                    "end": 392.12,
                    "word": " real",
                    "probability": 0.861328125
                },
                {
                    "start": 392.12,
                    "end": 392.44,
                    "word": " thing",
                    "probability": 0.79052734375
                },
                {
                    "start": 392.44,
                    "end": 393.0,
                    "word": ".",
                    "probability": 0.98193359375
                },
                {
                    "start": 393.14,
                    "end": 393.7,
                    "word": " For",
                    "probability": 0.89697265625
                },
                {
                    "start": 393.7,
                    "end": 394.04,
                    "word": " example",
                    "probability": 0.931640625
                },
                {
                    "start": 394.04,
                    "end": 394.12,
                    "word": ",",
                    "probability": 0.96337890625
                }
            ]
        },
        {
            "id": 63,
            "seek": 41606,
            "start": 394.12,
            "end": 399.38,
            "text": " in the right side of the slide, you can see that in the left image, there's this reconstructed image,",
            "tokens": [
                50364,
                294,
                264,
                558,
                1252,
                295,
                264,
                4137,
                11,
                291,
                393,
                536,
                300,
                294,
                264,
                1411,
                3256,
                11,
                456,
                311,
                341,
                31499,
                292,
                3256,
                11,
                50620
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2061543370387992,
            "compression_ratio": 1.785425101214575,
            "no_speech_prob": 0.0006985664367675781,
            "words": [
                {
                    "start": 394.12,
                    "end": 394.28,
                    "word": " in",
                    "probability": 0.290283203125
                },
                {
                    "start": 394.28,
                    "end": 394.4,
                    "word": " the",
                    "probability": 0.81005859375
                },
                {
                    "start": 394.4,
                    "end": 394.68,
                    "word": " right",
                    "probability": 0.86474609375
                },
                {
                    "start": 394.68,
                    "end": 395.2,
                    "word": " side",
                    "probability": 0.81103515625
                },
                {
                    "start": 395.2,
                    "end": 395.36,
                    "word": " of",
                    "probability": 0.90576171875
                },
                {
                    "start": 395.36,
                    "end": 395.62,
                    "word": " the",
                    "probability": 0.8095703125
                },
                {
                    "start": 395.62,
                    "end": 395.92,
                    "word": " slide",
                    "probability": 0.896484375
                },
                {
                    "start": 395.92,
                    "end": 396.0,
                    "word": ",",
                    "probability": 0.5478515625
                },
                {
                    "start": 396.0,
                    "end": 396.08,
                    "word": " you",
                    "probability": 0.88623046875
                },
                {
                    "start": 396.08,
                    "end": 396.22,
                    "word": " can",
                    "probability": 0.7978515625
                },
                {
                    "start": 396.22,
                    "end": 396.42,
                    "word": " see",
                    "probability": 0.8203125
                },
                {
                    "start": 396.42,
                    "end": 396.64,
                    "word": " that",
                    "probability": 0.873046875
                },
                {
                    "start": 396.64,
                    "end": 396.76,
                    "word": " in",
                    "probability": 0.88623046875
                },
                {
                    "start": 396.76,
                    "end": 396.88,
                    "word": " the",
                    "probability": 0.7744140625
                },
                {
                    "start": 396.88,
                    "end": 397.04,
                    "word": " left",
                    "probability": 0.85107421875
                },
                {
                    "start": 397.04,
                    "end": 397.48,
                    "word": " image",
                    "probability": 0.83349609375
                },
                {
                    "start": 397.48,
                    "end": 397.7,
                    "word": ",",
                    "probability": 0.642578125
                },
                {
                    "start": 397.7,
                    "end": 397.88,
                    "word": " there's",
                    "probability": 0.910400390625
                },
                {
                    "start": 397.88,
                    "end": 398.1,
                    "word": " this",
                    "probability": 0.8681640625
                },
                {
                    "start": 398.1,
                    "end": 398.68,
                    "word": " reconstructed",
                    "probability": 0.86669921875
                },
                {
                    "start": 398.68,
                    "end": 399.0,
                    "word": " image",
                    "probability": 0.8291015625
                },
                {
                    "start": 399.0,
                    "end": 399.38,
                    "word": ",",
                    "probability": 0.427490234375
                }
            ]
        },
        {
            "id": 64,
            "seek": 41606,
            "start": 399.38,
            "end": 403.66,
            "text": " and in the right is the original image, and they're very similar to each other.",
            "tokens": [
                50620,
                293,
                294,
                264,
                558,
                307,
                264,
                3380,
                3256,
                11,
                293,
                436,
                434,
                588,
                2531,
                281,
                1184,
                661,
                13,
                50816
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2061543370387992,
            "compression_ratio": 1.785425101214575,
            "no_speech_prob": 0.0006985664367675781,
            "words": [
                {
                    "start": 399.38,
                    "end": 399.48,
                    "word": " and",
                    "probability": 0.892578125
                },
                {
                    "start": 399.48,
                    "end": 399.56,
                    "word": " in",
                    "probability": 0.505859375
                },
                {
                    "start": 399.56,
                    "end": 399.7,
                    "word": " the",
                    "probability": 0.81103515625
                },
                {
                    "start": 399.7,
                    "end": 399.92,
                    "word": " right",
                    "probability": 0.87451171875
                },
                {
                    "start": 399.92,
                    "end": 400.38,
                    "word": " is",
                    "probability": 0.87109375
                },
                {
                    "start": 400.38,
                    "end": 400.68,
                    "word": " the",
                    "probability": 0.779296875
                },
                {
                    "start": 400.68,
                    "end": 401.38,
                    "word": " original",
                    "probability": 0.830078125
                },
                {
                    "start": 401.38,
                    "end": 401.78,
                    "word": " image",
                    "probability": 0.87060546875
                },
                {
                    "start": 401.78,
                    "end": 401.88,
                    "word": ",",
                    "probability": 0.460205078125
                },
                {
                    "start": 401.88,
                    "end": 401.96,
                    "word": " and",
                    "probability": 0.87548828125
                },
                {
                    "start": 401.96,
                    "end": 402.2,
                    "word": " they're",
                    "probability": 0.832763671875
                },
                {
                    "start": 402.2,
                    "end": 402.44,
                    "word": " very",
                    "probability": 0.7431640625
                },
                {
                    "start": 402.44,
                    "end": 402.64,
                    "word": " similar",
                    "probability": 0.9013671875
                },
                {
                    "start": 402.64,
                    "end": 402.8,
                    "word": " to",
                    "probability": 0.908203125
                },
                {
                    "start": 402.8,
                    "end": 402.94,
                    "word": " each",
                    "probability": 0.8681640625
                },
                {
                    "start": 402.94,
                    "end": 403.14,
                    "word": " other",
                    "probability": 0.7841796875
                },
                {
                    "start": 403.14,
                    "end": 403.66,
                    "word": ".",
                    "probability": 0.99169921875
                }
            ]
        },
        {
            "id": 65,
            "seek": 41606,
            "start": 404.18,
            "end": 408.86,
            "text": " In the next slide, we also have a membership inference attack, where an adversary can",
            "tokens": [
                50864,
                682,
                264,
                958,
                4137,
                11,
                321,
                611,
                362,
                257,
                16560,
                38253,
                2690,
                11,
                689,
                364,
                48222,
                393,
                51124
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2061543370387992,
            "compression_ratio": 1.785425101214575,
            "no_speech_prob": 0.0006985664367675781,
            "words": [
                {
                    "start": 404.18,
                    "end": 404.34,
                    "word": " In",
                    "probability": 0.72119140625
                },
                {
                    "start": 404.34,
                    "end": 404.44,
                    "word": " the",
                    "probability": 0.80615234375
                },
                {
                    "start": 404.44,
                    "end": 404.64,
                    "word": " next",
                    "probability": 0.85302734375
                },
                {
                    "start": 404.64,
                    "end": 404.96,
                    "word": " slide",
                    "probability": 0.89501953125
                },
                {
                    "start": 404.96,
                    "end": 405.08,
                    "word": ",",
                    "probability": 0.9052734375
                },
                {
                    "start": 405.08,
                    "end": 405.22,
                    "word": " we",
                    "probability": 0.89990234375
                },
                {
                    "start": 405.22,
                    "end": 405.46,
                    "word": " also",
                    "probability": 0.82470703125
                },
                {
                    "start": 405.46,
                    "end": 405.86,
                    "word": " have",
                    "probability": 0.88330078125
                },
                {
                    "start": 406.28,
                    "end": 406.6,
                    "word": " a membership",
                    "probability": 0.71826171875
                },
                {
                    "start": 406.6,
                    "end": 407.12,
                    "word": " inference",
                    "probability": 0.73583984375
                },
                {
                    "start": 407.12,
                    "end": 407.52,
                    "word": " attack",
                    "probability": 0.81298828125
                },
                {
                    "start": 407.52,
                    "end": 407.72,
                    "word": ",",
                    "probability": 0.427001953125
                },
                {
                    "start": 407.72,
                    "end": 407.86,
                    "word": " where",
                    "probability": 0.87109375
                },
                {
                    "start": 407.86,
                    "end": 408.04,
                    "word": " an",
                    "probability": 0.908203125
                },
                {
                    "start": 408.04,
                    "end": 408.48,
                    "word": " adversary",
                    "probability": 0.76318359375
                },
                {
                    "start": 408.48,
                    "end": 408.86,
                    "word": " can",
                    "probability": 0.8984375
                }
            ]
        },
        {
            "id": 66,
            "seek": 41606,
            "start": 408.86,
            "end": 413.92,
            "text": " generate multiple shadow network models that resembles the production environment,",
            "tokens": [
                51124,
                8460,
                3866,
                8576,
                3209,
                5245,
                300,
                34433,
                264,
                4265,
                2823,
                11,
                51348
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2061543370387992,
            "compression_ratio": 1.785425101214575,
            "no_speech_prob": 0.0006985664367675781,
            "words": [
                {
                    "start": 408.86,
                    "end": 409.76,
                    "word": " generate",
                    "probability": 0.8212890625
                },
                {
                    "start": 409.76,
                    "end": 410.48,
                    "word": " multiple",
                    "probability": 0.791015625
                },
                {
                    "start": 410.48,
                    "end": 410.86,
                    "word": " shadow",
                    "probability": 0.7138671875
                },
                {
                    "start": 410.86,
                    "end": 411.16,
                    "word": " network",
                    "probability": 0.83056640625
                },
                {
                    "start": 411.16,
                    "end": 411.48,
                    "word": " models",
                    "probability": 0.77978515625
                },
                {
                    "start": 411.48,
                    "end": 411.76,
                    "word": " that",
                    "probability": 0.86572265625
                },
                {
                    "start": 411.76,
                    "end": 412.16,
                    "word": " resembles",
                    "probability": 0.5146484375
                },
                {
                    "start": 412.16,
                    "end": 412.52,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 412.52,
                    "end": 412.76,
                    "word": " production",
                    "probability": 0.794921875
                },
                {
                    "start": 412.76,
                    "end": 413.3,
                    "word": " environment",
                    "probability": 0.6533203125
                },
                {
                    "start": 413.3,
                    "end": 413.92,
                    "word": ",",
                    "probability": 0.9208984375
                }
            ]
        },
        {
            "id": 67,
            "seek": 41606,
            "start": 413.92,
            "end": 419.0,
            "text": " produce the probabilities from these models, create a new model called the attack model to",
            "tokens": [
                51348,
                5258,
                264,
                33783,
                490,
                613,
                5245,
                11,
                1884,
                257,
                777,
                2316,
                1219,
                264,
                2690,
                2316,
                281,
                51616
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2061543370387992,
            "compression_ratio": 1.785425101214575,
            "no_speech_prob": 0.0006985664367675781,
            "words": [
                {
                    "start": 413.92,
                    "end": 414.2,
                    "word": " produce",
                    "probability": 0.6826171875
                },
                {
                    "start": 414.2,
                    "end": 414.38,
                    "word": " the",
                    "probability": 0.509765625
                },
                {
                    "start": 414.38,
                    "end": 414.82,
                    "word": " probabilities",
                    "probability": 0.8310546875
                },
                {
                    "start": 414.82,
                    "end": 415.24,
                    "word": " from",
                    "probability": 0.822265625
                },
                {
                    "start": 415.24,
                    "end": 415.48,
                    "word": " these",
                    "probability": 0.6943359375
                },
                {
                    "start": 415.48,
                    "end": 415.86,
                    "word": " models",
                    "probability": 0.78466796875
                },
                {
                    "start": 415.86,
                    "end": 416.5,
                    "word": ",",
                    "probability": 0.98876953125
                },
                {
                    "start": 416.5,
                    "end": 416.7,
                    "word": " create",
                    "probability": 0.82861328125
                },
                {
                    "start": 416.96,
                    "end": 417.2,
                    "word": " a new",
                    "probability": 0.81640625
                },
                {
                    "start": 417.2,
                    "end": 417.52,
                    "word": " model",
                    "probability": 0.8193359375
                },
                {
                    "start": 417.52,
                    "end": 417.88,
                    "word": " called",
                    "probability": 0.6708984375
                },
                {
                    "start": 417.88,
                    "end": 418.0,
                    "word": " the",
                    "probability": 0.79638671875
                },
                {
                    "start": 418.0,
                    "end": 418.34,
                    "word": " attack",
                    "probability": 0.68310546875
                },
                {
                    "start": 418.34,
                    "end": 418.68,
                    "word": " model",
                    "probability": 0.8251953125
                },
                {
                    "start": 418.68,
                    "end": 419.0,
                    "word": " to",
                    "probability": 0.8603515625
                }
            ]
        },
        {
            "id": 68,
            "seek": 44554,
            "start": 419.0,
            "end": 425.4,
            "text": " feed in these probabilities, and learn what observations are trained by the production model",
            "tokens": [
                50368,
                3154,
                294,
                613,
                33783,
                11,
                293,
                1466,
                437,
                18163,
                366,
                8895,
                538,
                264,
                4265,
                2316,
                50684
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21857829703079476,
            "compression_ratio": 1.7568807339449541,
            "no_speech_prob": 0.0008039474487304688,
            "words": [
                {
                    "start": 419.0,
                    "end": 419.38,
                    "word": " feed",
                    "probability": 0.1268310546875
                },
                {
                    "start": 419.38,
                    "end": 419.58,
                    "word": " in",
                    "probability": 0.88427734375
                },
                {
                    "start": 419.58,
                    "end": 419.76,
                    "word": " these",
                    "probability": 0.697265625
                },
                {
                    "start": 419.76,
                    "end": 420.22,
                    "word": " probabilities",
                    "probability": 0.80322265625
                },
                {
                    "start": 420.22,
                    "end": 420.86,
                    "word": ",",
                    "probability": 0.1082763671875
                },
                {
                    "start": 420.86,
                    "end": 420.92,
                    "word": " and",
                    "probability": 0.8857421875
                },
                {
                    "start": 420.92,
                    "end": 421.24,
                    "word": " learn",
                    "probability": 0.7763671875
                },
                {
                    "start": 421.24,
                    "end": 421.86,
                    "word": " what",
                    "probability": 0.8896484375
                },
                {
                    "start": 421.86,
                    "end": 422.62,
                    "word": " observations",
                    "probability": 0.6923828125
                },
                {
                    "start": 422.62,
                    "end": 423.4,
                    "word": " are",
                    "probability": 0.89599609375
                },
                {
                    "start": 423.4,
                    "end": 424.52,
                    "word": " trained",
                    "probability": 0.66650390625
                },
                {
                    "start": 424.52,
                    "end": 424.76,
                    "word": " by",
                    "probability": 0.90283203125
                },
                {
                    "start": 424.76,
                    "end": 424.92,
                    "word": " the",
                    "probability": 0.7470703125
                },
                {
                    "start": 424.92,
                    "end": 425.16,
                    "word": " production",
                    "probability": 0.8212890625
                },
                {
                    "start": 425.16,
                    "end": 425.4,
                    "word": " model",
                    "probability": 0.76904296875
                }
            ]
        },
        {
            "id": 69,
            "seek": 44554,
            "start": 425.4,
            "end": 429.54,
            "text": " and what are not trained by the production model, identify what are the members and what are not the",
            "tokens": [
                50684,
                293,
                437,
                366,
                406,
                8895,
                538,
                264,
                4265,
                2316,
                11,
                5876,
                437,
                366,
                264,
                2679,
                293,
                437,
                366,
                406,
                264,
                50884
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21857829703079476,
            "compression_ratio": 1.7568807339449541,
            "no_speech_prob": 0.0008039474487304688,
            "words": [
                {
                    "start": 425.4,
                    "end": 425.64,
                    "word": " and",
                    "probability": 0.70947265625
                },
                {
                    "start": 425.64,
                    "end": 425.96,
                    "word": " what",
                    "probability": 0.88232421875
                },
                {
                    "start": 425.96,
                    "end": 426.28,
                    "word": " are",
                    "probability": 0.8369140625
                },
                {
                    "start": 426.28,
                    "end": 426.48,
                    "word": " not",
                    "probability": 0.8896484375
                },
                {
                    "start": 426.48,
                    "end": 426.72,
                    "word": " trained",
                    "probability": 0.83203125
                },
                {
                    "start": 426.72,
                    "end": 426.88,
                    "word": " by",
                    "probability": 0.91845703125
                },
                {
                    "start": 426.88,
                    "end": 427.0,
                    "word": " the",
                    "probability": 0.7822265625
                },
                {
                    "start": 427.0,
                    "end": 427.2,
                    "word": " production",
                    "probability": 0.87939453125
                },
                {
                    "start": 427.2,
                    "end": 427.42,
                    "word": " model",
                    "probability": 0.82861328125
                },
                {
                    "start": 427.42,
                    "end": 427.7,
                    "word": ",",
                    "probability": 0.367919921875
                },
                {
                    "start": 427.7,
                    "end": 428.0,
                    "word": " identify",
                    "probability": 0.68896484375
                },
                {
                    "start": 428.0,
                    "end": 428.24,
                    "word": " what",
                    "probability": 0.875
                },
                {
                    "start": 428.24,
                    "end": 428.36,
                    "word": " are",
                    "probability": 0.8974609375
                },
                {
                    "start": 428.36,
                    "end": 428.44,
                    "word": " the",
                    "probability": 0.8017578125
                },
                {
                    "start": 428.44,
                    "end": 428.68,
                    "word": " members",
                    "probability": 0.501953125
                },
                {
                    "start": 428.68,
                    "end": 428.88,
                    "word": " and",
                    "probability": 0.74560546875
                },
                {
                    "start": 428.88,
                    "end": 429.0,
                    "word": " what",
                    "probability": 0.859375
                },
                {
                    "start": 429.0,
                    "end": 429.14,
                    "word": " are",
                    "probability": 0.876953125
                },
                {
                    "start": 429.14,
                    "end": 429.34,
                    "word": " not",
                    "probability": 0.88525390625
                },
                {
                    "start": 429.34,
                    "end": 429.54,
                    "word": " the",
                    "probability": 0.78515625
                }
            ]
        },
        {
            "id": 70,
            "seek": 44554,
            "start": 429.54,
            "end": 438.38,
            "text": " members. In the next slide, we also have a core attack. This is more prone to models such as LSTM,",
            "tokens": [
                50884,
                2679,
                13,
                682,
                264,
                958,
                4137,
                11,
                321,
                611,
                362,
                257,
                4965,
                2690,
                13,
                639,
                307,
                544,
                25806,
                281,
                5245,
                1270,
                382,
                441,
                6840,
                44,
                11,
                51292
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21857829703079476,
            "compression_ratio": 1.7568807339449541,
            "no_speech_prob": 0.0008039474487304688,
            "words": [
                {
                    "start": 429.54,
                    "end": 429.88,
                    "word": " members",
                    "probability": 0.76953125
                },
                {
                    "start": 429.88,
                    "end": 430.4,
                    "word": ".",
                    "probability": 0.95263671875
                },
                {
                    "start": 430.9,
                    "end": 431.42,
                    "word": " In",
                    "probability": 0.58251953125
                },
                {
                    "start": 431.42,
                    "end": 431.52,
                    "word": " the",
                    "probability": 0.81103515625
                },
                {
                    "start": 431.52,
                    "end": 431.7,
                    "word": " next",
                    "probability": 0.85498046875
                },
                {
                    "start": 431.7,
                    "end": 431.9,
                    "word": " slide",
                    "probability": 0.8916015625
                },
                {
                    "start": 431.9,
                    "end": 431.96,
                    "word": ",",
                    "probability": 0.626953125
                },
                {
                    "start": 431.96,
                    "end": 432.06,
                    "word": " we",
                    "probability": 0.90478515625
                },
                {
                    "start": 432.06,
                    "end": 432.38,
                    "word": " also",
                    "probability": 0.82470703125
                },
                {
                    "start": 432.38,
                    "end": 432.96,
                    "word": " have",
                    "probability": 0.88037109375
                },
                {
                    "start": 433.44,
                    "end": 433.62,
                    "word": " a core",
                    "probability": 0.299560546875
                },
                {
                    "start": 433.62,
                    "end": 434.0,
                    "word": " attack",
                    "probability": 0.599609375
                },
                {
                    "start": 434.0,
                    "end": 434.46,
                    "word": ".",
                    "probability": 0.849609375
                },
                {
                    "start": 434.46,
                    "end": 434.86,
                    "word": " This",
                    "probability": 0.794921875
                },
                {
                    "start": 434.86,
                    "end": 435.32,
                    "word": " is",
                    "probability": 0.9091796875
                },
                {
                    "start": 435.32,
                    "end": 435.64,
                    "word": " more",
                    "probability": 0.880859375
                },
                {
                    "start": 435.64,
                    "end": 435.9,
                    "word": " prone",
                    "probability": 0.4990234375
                },
                {
                    "start": 435.9,
                    "end": 436.24,
                    "word": " to",
                    "probability": 0.908203125
                },
                {
                    "start": 436.24,
                    "end": 436.62,
                    "word": " models",
                    "probability": 0.802734375
                },
                {
                    "start": 436.62,
                    "end": 436.82,
                    "word": " such",
                    "probability": 0.86962890625
                },
                {
                    "start": 436.82,
                    "end": 436.98,
                    "word": " as",
                    "probability": 0.90380859375
                },
                {
                    "start": 436.98,
                    "end": 437.74,
                    "word": " LSTM",
                    "probability": 0.9412434895833334
                },
                {
                    "start": 437.74,
                    "end": 438.38,
                    "word": ",",
                    "probability": 0.474853515625
                }
            ]
        },
        {
            "id": 71,
            "seek": 44554,
            "start": 438.38,
            "end": 448.48,
            "text": " or models that learn from a time series of data points. For this example, you might have a",
            "tokens": [
                51324,
                420,
                5245,
                300,
                1466,
                490,
                257,
                565,
                2638,
                295,
                1412,
                2793,
                13,
                1171,
                341,
                1365,
                11,
                291,
                1062,
                362,
                257,
                51720
            ],
            "temperature": 0.0,
            "avg_logprob": -0.21857829703079476,
            "compression_ratio": 1.7568807339449541,
            "no_speech_prob": 0.0008039474487304688,
            "words": [
                {
                    "start": 438.38,
                    "end": 438.8,
                    "word": " or",
                    "probability": 0.9052734375
                },
                {
                    "start": 438.8,
                    "end": 440.22,
                    "word": " models",
                    "probability": 0.74560546875
                },
                {
                    "start": 440.22,
                    "end": 440.62,
                    "word": " that",
                    "probability": 0.88232421875
                },
                {
                    "start": 440.62,
                    "end": 441.6,
                    "word": " learn",
                    "probability": 0.78564453125
                },
                {
                    "start": 441.6,
                    "end": 441.84,
                    "word": " from",
                    "probability": 0.81689453125
                },
                {
                    "start": 441.98,
                    "end": 442.16,
                    "word": " a time",
                    "probability": 0.83740234375
                },
                {
                    "start": 442.16,
                    "end": 442.46,
                    "word": " series",
                    "probability": 0.81591796875
                },
                {
                    "start": 442.46,
                    "end": 442.7,
                    "word": " of",
                    "probability": 0.91064453125
                },
                {
                    "start": 442.7,
                    "end": 442.84,
                    "word": " data",
                    "probability": 0.78564453125
                },
                {
                    "start": 442.84,
                    "end": 443.16,
                    "word": " points",
                    "probability": 0.783203125
                },
                {
                    "start": 443.16,
                    "end": 443.68,
                    "word": ".",
                    "probability": 0.9794921875
                },
                {
                    "start": 443.74,
                    "end": 444.26,
                    "word": " For",
                    "probability": 0.9052734375
                },
                {
                    "start": 444.26,
                    "end": 444.72,
                    "word": " this",
                    "probability": 0.880859375
                },
                {
                    "start": 444.72,
                    "end": 445.32,
                    "word": " example",
                    "probability": 0.92578125
                },
                {
                    "start": 445.32,
                    "end": 445.68,
                    "word": ",",
                    "probability": 0.97021484375
                },
                {
                    "start": 445.68,
                    "end": 445.84,
                    "word": " you",
                    "probability": 0.88037109375
                },
                {
                    "start": 445.84,
                    "end": 446.0,
                    "word": " might",
                    "probability": 0.78076171875
                },
                {
                    "start": 446.0,
                    "end": 446.16,
                    "word": " have",
                    "probability": 0.88037109375
                },
                {
                    "start": 446.16,
                    "end": 448.48,
                    "word": " a",
                    "probability": 0.99658203125
                }
            ]
        },
        {
            "id": 72,
            "seek": 47004,
            "start": 448.48,
            "end": 452.02,
            "text": " model that learns from social security, and it does an autocomplete, and one of the autocomplete",
            "tokens": [
                50364,
                2316,
                300,
                27152,
                490,
                2093,
                3825,
                11,
                293,
                309,
                775,
                364,
                45833,
                298,
                17220,
                11,
                293,
                472,
                295,
                264,
                45833,
                298,
                17220,
                50604
            ],
            "temperature": 0.0,
            "avg_logprob": -0.36203458302832664,
            "compression_ratio": 1.6234309623430963,
            "no_speech_prob": 0.01495361328125,
            "words": [
                {
                    "start": 448.48,
                    "end": 448.48,
                    "word": " model",
                    "probability": 0.0007195472717285156
                },
                {
                    "start": 448.48,
                    "end": 448.48,
                    "word": " that",
                    "probability": 0.07415771484375
                },
                {
                    "start": 448.48,
                    "end": 448.48,
                    "word": " learns",
                    "probability": 0.00042438507080078125
                },
                {
                    "start": 448.48,
                    "end": 448.48,
                    "word": " from",
                    "probability": 0.130859375
                },
                {
                    "start": 448.48,
                    "end": 448.64,
                    "word": " social",
                    "probability": 0.6015625
                },
                {
                    "start": 448.64,
                    "end": 449.04,
                    "word": " security",
                    "probability": 0.9296875
                },
                {
                    "start": 449.04,
                    "end": 449.16,
                    "word": ",",
                    "probability": 0.13720703125
                },
                {
                    "start": 449.16,
                    "end": 449.24,
                    "word": " and",
                    "probability": 0.8515625
                },
                {
                    "start": 449.24,
                    "end": 449.34,
                    "word": " it",
                    "probability": 0.845703125
                },
                {
                    "start": 449.34,
                    "end": 449.48,
                    "word": " does",
                    "probability": 0.8701171875
                },
                {
                    "start": 449.48,
                    "end": 449.64,
                    "word": " an",
                    "probability": 0.52099609375
                },
                {
                    "start": 449.64,
                    "end": 450.28,
                    "word": " autocomplete",
                    "probability": 0.7903645833333334
                },
                {
                    "start": 450.28,
                    "end": 450.78,
                    "word": ",",
                    "probability": 0.341064453125
                },
                {
                    "start": 450.78,
                    "end": 450.9,
                    "word": " and",
                    "probability": 0.8798828125
                },
                {
                    "start": 450.9,
                    "end": 451.16,
                    "word": " one",
                    "probability": 0.83740234375
                },
                {
                    "start": 451.16,
                    "end": 451.3,
                    "word": " of",
                    "probability": 0.90283203125
                },
                {
                    "start": 451.3,
                    "end": 451.42,
                    "word": " the",
                    "probability": 0.8125
                },
                {
                    "start": 451.42,
                    "end": 452.02,
                    "word": " autocomplete",
                    "probability": 0.8116861979166666
                }
            ]
        },
        {
            "id": 73,
            "seek": 47004,
            "start": 452.02,
            "end": 458.1,
            "text": " could be some sort of sensitive information, such as the phone number or the social security of a",
            "tokens": [
                50604,
                727,
                312,
                512,
                1333,
                295,
                9477,
                1589,
                11,
                1270,
                382,
                264,
                2593,
                1230,
                420,
                264,
                2093,
                3825,
                295,
                257,
                50880
            ],
            "temperature": 0.0,
            "avg_logprob": -0.36203458302832664,
            "compression_ratio": 1.6234309623430963,
            "no_speech_prob": 0.01495361328125,
            "words": [
                {
                    "start": 452.02,
                    "end": 452.3,
                    "word": " could",
                    "probability": 0.79345703125
                },
                {
                    "start": 452.3,
                    "end": 452.5,
                    "word": " be",
                    "probability": 0.90625
                },
                {
                    "start": 452.5,
                    "end": 452.7,
                    "word": " some",
                    "probability": 0.81884765625
                },
                {
                    "start": 452.7,
                    "end": 452.88,
                    "word": " sort",
                    "probability": 0.7607421875
                },
                {
                    "start": 452.88,
                    "end": 453.0,
                    "word": " of",
                    "probability": 0.90771484375
                },
                {
                    "start": 453.0,
                    "end": 453.22,
                    "word": " sensitive",
                    "probability": 0.69677734375
                },
                {
                    "start": 453.22,
                    "end": 453.88,
                    "word": " information",
                    "probability": 0.73291015625
                },
                {
                    "start": 453.88,
                    "end": 454.08,
                    "word": ",",
                    "probability": 0.343017578125
                },
                {
                    "start": 454.08,
                    "end": 454.22,
                    "word": " such",
                    "probability": 0.892578125
                },
                {
                    "start": 454.22,
                    "end": 454.58,
                    "word": " as",
                    "probability": 0.90380859375
                },
                {
                    "start": 454.58,
                    "end": 455.38,
                    "word": " the",
                    "probability": 0.7958984375
                },
                {
                    "start": 455.38,
                    "end": 455.6,
                    "word": " phone",
                    "probability": 0.87353515625
                },
                {
                    "start": 455.6,
                    "end": 455.88,
                    "word": " number",
                    "probability": 0.8740234375
                },
                {
                    "start": 455.88,
                    "end": 456.24,
                    "word": " or",
                    "probability": 0.80126953125
                },
                {
                    "start": 456.24,
                    "end": 456.66,
                    "word": " the",
                    "probability": 0.7509765625
                },
                {
                    "start": 456.66,
                    "end": 457.08,
                    "word": " social",
                    "probability": 0.77001953125
                },
                {
                    "start": 457.08,
                    "end": 457.4,
                    "word": " security",
                    "probability": 0.91796875
                },
                {
                    "start": 457.4,
                    "end": 457.58,
                    "word": " of",
                    "probability": 0.89599609375
                },
                {
                    "start": 457.72,
                    "end": 458.1,
                    "word": " a person",
                    "probability": 0.79150390625
                }
            ]
        },
        {
            "id": 74,
            "seek": 47004,
            "start": 458.1,
            "end": 466.34,
            "text": " person. This is more prone to maybe like if you're over-memorizing from the neural network model",
            "tokens": [
                50880,
                954,
                13,
                639,
                307,
                544,
                25806,
                281,
                1310,
                411,
                498,
                291,
                434,
                670,
                12,
                17886,
                284,
                3319,
                490,
                264,
                18161,
                3209,
                2316,
                51292
            ],
            "temperature": 0.0,
            "avg_logprob": -0.36203458302832664,
            "compression_ratio": 1.6234309623430963,
            "no_speech_prob": 0.01495361328125,
            "words": [
                {
                    "start": 458.1,
                    "end": 458.66,
                    "word": ".",
                    "probability": 0.9619140625
                },
                {
                    "start": 458.92,
                    "end": 459.2,
                    "word": " This",
                    "probability": 0.79736328125
                },
                {
                    "start": 459.2,
                    "end": 459.52,
                    "word": " is",
                    "probability": 0.90869140625
                },
                {
                    "start": 459.52,
                    "end": 459.86,
                    "word": " more",
                    "probability": 0.87890625
                },
                {
                    "start": 459.86,
                    "end": 460.12,
                    "word": " prone",
                    "probability": 0.67626953125
                },
                {
                    "start": 460.12,
                    "end": 460.6,
                    "word": " to",
                    "probability": 0.90234375
                },
                {
                    "start": 460.6,
                    "end": 461.72,
                    "word": " maybe",
                    "probability": 0.468994140625
                },
                {
                    "start": 461.72,
                    "end": 462.04,
                    "word": " like",
                    "probability": 0.41064453125
                },
                {
                    "start": 462.04,
                    "end": 462.38,
                    "word": " if",
                    "probability": 0.72998046875
                },
                {
                    "start": 462.38,
                    "end": 463.78,
                    "word": " you're",
                    "probability": 0.74072265625
                },
                {
                    "start": 463.78,
                    "end": 464.2,
                    "word": " over",
                    "probability": 0.8974609375
                },
                {
                    "start": 464.2,
                    "end": 464.66,
                    "word": "-memorizing",
                    "probability": 0.7685546875
                },
                {
                    "start": 464.66,
                    "end": 464.94,
                    "word": " from",
                    "probability": 0.802734375
                },
                {
                    "start": 464.94,
                    "end": 465.1,
                    "word": " the",
                    "probability": 0.78125
                },
                {
                    "start": 465.1,
                    "end": 465.24,
                    "word": " neural",
                    "probability": 0.89501953125
                },
                {
                    "start": 465.24,
                    "end": 465.6,
                    "word": " network",
                    "probability": 0.86279296875
                },
                {
                    "start": 465.6,
                    "end": 465.84,
                    "word": " model",
                    "probability": 0.76904296875
                },
                {
                    "start": 465.84,
                    "end": 466.34,
                    "word": " itself",
                    "probability": 0.76416015625
                }
            ]
        },
        {
            "id": 75,
            "seek": 47004,
            "start": 466.34,
            "end": 472.98,
            "text": " itself, where the records might appear at least five times. In the next slide, we have a summary",
            "tokens": [
                51292,
                2564,
                11,
                689,
                264,
                7724,
                1062,
                4204,
                412,
                1935,
                1732,
                1413,
                13,
                682,
                264,
                958,
                4137,
                11,
                321,
                362,
                257,
                12691,
                51652
            ],
            "temperature": 0.0,
            "avg_logprob": -0.36203458302832664,
            "compression_ratio": 1.6234309623430963,
            "no_speech_prob": 0.01495361328125,
            "words": [
                {
                    "start": 466.34,
                    "end": 466.84,
                    "word": ",",
                    "probability": 0.8447265625
                },
                {
                    "start": 466.84,
                    "end": 467.3,
                    "word": " where",
                    "probability": 0.8876953125
                },
                {
                    "start": 467.3,
                    "end": 467.54,
                    "word": " the",
                    "probability": 0.79248046875
                },
                {
                    "start": 467.54,
                    "end": 467.78,
                    "word": " records",
                    "probability": 0.76904296875
                },
                {
                    "start": 467.78,
                    "end": 467.96,
                    "word": " might",
                    "probability": 0.67919921875
                },
                {
                    "start": 467.96,
                    "end": 468.24,
                    "word": " appear",
                    "probability": 0.7490234375
                },
                {
                    "start": 468.24,
                    "end": 468.64,
                    "word": " at",
                    "probability": 0.5517578125
                },
                {
                    "start": 468.64,
                    "end": 469.34,
                    "word": " least",
                    "probability": 0.8203125
                },
                {
                    "start": 469.34,
                    "end": 469.6,
                    "word": " five",
                    "probability": 0.763671875
                },
                {
                    "start": 469.6,
                    "end": 470.04,
                    "word": " times",
                    "probability": 0.83349609375
                },
                {
                    "start": 470.04,
                    "end": 470.6,
                    "word": ".",
                    "probability": 0.99462890625
                },
                {
                    "start": 471.04,
                    "end": 471.6,
                    "word": " In",
                    "probability": 0.51611328125
                },
                {
                    "start": 471.6,
                    "end": 471.68,
                    "word": " the",
                    "probability": 0.80712890625
                },
                {
                    "start": 471.68,
                    "end": 471.86,
                    "word": " next",
                    "probability": 0.857421875
                },
                {
                    "start": 471.86,
                    "end": 472.08,
                    "word": " slide",
                    "probability": 0.884765625
                },
                {
                    "start": 472.08,
                    "end": 472.14,
                    "word": ",",
                    "probability": 0.89892578125
                },
                {
                    "start": 472.14,
                    "end": 472.24,
                    "word": " we",
                    "probability": 0.90625
                },
                {
                    "start": 472.24,
                    "end": 472.54,
                    "word": " have",
                    "probability": 0.876953125
                },
                {
                    "start": 472.78,
                    "end": 472.98,
                    "word": " a summary",
                    "probability": 0.78173828125
                }
            ]
        },
        {
            "id": 76,
            "seek": 49438,
            "start": 472.98,
            "end": 478.84,
            "text": " of different privacy attacks that I've talked about earlier on. You could take a pause here",
            "tokens": [
                50364,
                295,
                819,
                11427,
                8122,
                300,
                286,
                600,
                2825,
                466,
                3071,
                322,
                13,
                509,
                727,
                747,
                257,
                10465,
                510,
                50652
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2826344929163969,
            "compression_ratio": 1.5529953917050692,
            "no_speech_prob": 0.0030269622802734375,
            "words": [
                {
                    "start": 472.98,
                    "end": 472.98,
                    "word": " of",
                    "probability": 0.060211181640625
                },
                {
                    "start": 472.98,
                    "end": 473.56,
                    "word": " different",
                    "probability": 0.229248046875
                },
                {
                    "start": 473.56,
                    "end": 474.08,
                    "word": " privacy",
                    "probability": 0.73583984375
                },
                {
                    "start": 474.08,
                    "end": 474.54,
                    "word": " attacks",
                    "probability": 0.279052734375
                },
                {
                    "start": 474.54,
                    "end": 475.34,
                    "word": " that",
                    "probability": 0.77734375
                },
                {
                    "start": 475.34,
                    "end": 475.98,
                    "word": " I've",
                    "probability": 0.5828857421875
                },
                {
                    "start": 475.98,
                    "end": 476.2,
                    "word": " talked",
                    "probability": 0.7626953125
                },
                {
                    "start": 476.2,
                    "end": 476.54,
                    "word": " about",
                    "probability": 0.82568359375
                },
                {
                    "start": 476.54,
                    "end": 476.9,
                    "word": " earlier",
                    "probability": 0.904296875
                },
                {
                    "start": 476.9,
                    "end": 477.18,
                    "word": " on",
                    "probability": 0.88232421875
                },
                {
                    "start": 477.18,
                    "end": 477.78,
                    "word": ".",
                    "probability": 0.8271484375
                },
                {
                    "start": 477.84,
                    "end": 477.92,
                    "word": " You",
                    "probability": 0.8544921875
                },
                {
                    "start": 477.92,
                    "end": 478.04,
                    "word": " could",
                    "probability": 0.261474609375
                },
                {
                    "start": 478.04,
                    "end": 478.18,
                    "word": " take",
                    "probability": 0.8134765625
                },
                {
                    "start": 478.32,
                    "end": 478.56,
                    "word": " a pause",
                    "probability": 0.787109375
                },
                {
                    "start": 478.56,
                    "end": 478.84,
                    "word": " here",
                    "probability": 0.81103515625
                }
            ]
        },
        {
            "id": 77,
            "seek": 49438,
            "start": 478.84,
            "end": 484.36,
            "text": " and just look through the difference, compare the different approaches, and look at the examples,",
            "tokens": [
                50652,
                293,
                445,
                574,
                807,
                264,
                2649,
                11,
                6794,
                264,
                819,
                11587,
                11,
                293,
                574,
                412,
                264,
                5110,
                11,
                50912
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2826344929163969,
            "compression_ratio": 1.5529953917050692,
            "no_speech_prob": 0.0030269622802734375,
            "words": [
                {
                    "start": 478.84,
                    "end": 479.02,
                    "word": " and",
                    "probability": 0.8671875
                },
                {
                    "start": 479.02,
                    "end": 479.2,
                    "word": " just",
                    "probability": 0.84033203125
                },
                {
                    "start": 479.2,
                    "end": 479.38,
                    "word": " look",
                    "probability": 0.7587890625
                },
                {
                    "start": 479.38,
                    "end": 479.74,
                    "word": " through",
                    "probability": 0.75341796875
                },
                {
                    "start": 479.74,
                    "end": 480.24,
                    "word": " the",
                    "probability": 0.66845703125
                },
                {
                    "start": 480.24,
                    "end": 480.72,
                    "word": " difference",
                    "probability": 0.2388916015625
                },
                {
                    "start": 480.72,
                    "end": 480.96,
                    "word": ",",
                    "probability": 0.572265625
                },
                {
                    "start": 480.96,
                    "end": 481.2,
                    "word": " compare",
                    "probability": 0.76513671875
                },
                {
                    "start": 481.2,
                    "end": 481.32,
                    "word": " the",
                    "probability": 0.7275390625
                },
                {
                    "start": 481.32,
                    "end": 481.64,
                    "word": " different",
                    "probability": 0.6787109375
                },
                {
                    "start": 481.64,
                    "end": 482.16,
                    "word": " approaches",
                    "probability": 0.68994140625
                },
                {
                    "start": 482.16,
                    "end": 482.6,
                    "word": ",",
                    "probability": 0.305908203125
                },
                {
                    "start": 482.6,
                    "end": 483.06,
                    "word": " and",
                    "probability": 0.8896484375
                },
                {
                    "start": 483.06,
                    "end": 483.24,
                    "word": " look",
                    "probability": 0.86328125
                },
                {
                    "start": 483.24,
                    "end": 483.34,
                    "word": " at",
                    "probability": 0.90869140625
                },
                {
                    "start": 483.34,
                    "end": 483.52,
                    "word": " the",
                    "probability": 0.8037109375
                },
                {
                    "start": 483.52,
                    "end": 483.98,
                    "word": " examples",
                    "probability": 0.78076171875
                },
                {
                    "start": 483.98,
                    "end": 484.36,
                    "word": ",",
                    "probability": 0.442138671875
                }
            ]
        },
        {
            "id": 78,
            "seek": 49438,
            "start": 484.36,
            "end": 490.95,
            "text": " but we can move on if you're ready to move on. Next slide.",
            "tokens": [
                50944,
                457,
                321,
                393,
                1286,
                322,
                498,
                291,
                434,
                1919,
                281,
                1286,
                322,
                13,
                3087,
                4137,
                13,
                51160
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2826344929163969,
            "compression_ratio": 1.5529953917050692,
            "no_speech_prob": 0.0030269622802734375,
            "words": [
                {
                    "start": 484.36,
                    "end": 484.88,
                    "word": " but",
                    "probability": 0.7724609375
                },
                {
                    "start": 484.88,
                    "end": 485.0,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 485.0,
                    "end": 485.1,
                    "word": " can",
                    "probability": 0.8779296875
                },
                {
                    "start": 485.1,
                    "end": 485.3,
                    "word": " move",
                    "probability": 0.8837890625
                },
                {
                    "start": 485.3,
                    "end": 485.6,
                    "word": " on",
                    "probability": 0.90576171875
                },
                {
                    "start": 485.6,
                    "end": 486.18,
                    "word": " if",
                    "probability": 0.837890625
                },
                {
                    "start": 486.18,
                    "end": 486.34,
                    "word": " you're",
                    "probability": 0.9033203125
                },
                {
                    "start": 486.34,
                    "end": 486.5,
                    "word": " ready",
                    "probability": 0.83349609375
                },
                {
                    "start": 486.5,
                    "end": 486.62,
                    "word": " to",
                    "probability": 0.7939453125
                },
                {
                    "start": 486.62,
                    "end": 486.72,
                    "word": " move",
                    "probability": 0.73486328125
                },
                {
                    "start": 486.72,
                    "end": 487.0,
                    "word": " on",
                    "probability": 0.9072265625
                },
                {
                    "start": 487.0,
                    "end": 487.6,
                    "word": ".",
                    "probability": 0.98583984375
                },
                {
                    "start": 489.87,
                    "end": 490.11,
                    "word": " Next",
                    "probability": 0.80810546875
                },
                {
                    "start": 490.11,
                    "end": 490.35,
                    "word": " slide",
                    "probability": 0.88232421875
                },
                {
                    "start": 490.35,
                    "end": 490.95,
                    "word": ".",
                    "probability": 0.9443359375
                }
            ]
        },
        {
            "id": 79,
            "seek": 49438,
            "start": 491.83,
            "end": 498.69,
            "text": " All right, so now we're really talking about methods and techniques to preserve privacy.",
            "tokens": [
                51252,
                1057,
                558,
                11,
                370,
                586,
                321,
                434,
                534,
                1417,
                466,
                7150,
                293,
                7512,
                281,
                15665,
                11427,
                13,
                51568
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2826344929163969,
            "compression_ratio": 1.5529953917050692,
            "no_speech_prob": 0.0030269622802734375,
            "words": [
                {
                    "start": 491.83,
                    "end": 492.41,
                    "word": " All",
                    "probability": 0.45458984375
                },
                {
                    "start": 492.41,
                    "end": 492.59,
                    "word": " right",
                    "probability": 0.87939453125
                },
                {
                    "start": 492.59,
                    "end": 492.81,
                    "word": ",",
                    "probability": 0.44580078125
                },
                {
                    "start": 492.81,
                    "end": 493.31,
                    "word": " so",
                    "probability": 0.86376953125
                },
                {
                    "start": 493.31,
                    "end": 493.53,
                    "word": " now",
                    "probability": 0.87744140625
                },
                {
                    "start": 493.53,
                    "end": 493.75,
                    "word": " we're",
                    "probability": 0.6453857421875
                },
                {
                    "start": 493.75,
                    "end": 493.93,
                    "word": " really",
                    "probability": 0.49755859375
                },
                {
                    "start": 493.93,
                    "end": 494.35,
                    "word": " talking",
                    "probability": 0.77880859375
                },
                {
                    "start": 494.35,
                    "end": 495.07,
                    "word": " about",
                    "probability": 0.82275390625
                },
                {
                    "start": 495.07,
                    "end": 496.17,
                    "word": " methods",
                    "probability": 0.8203125
                },
                {
                    "start": 496.17,
                    "end": 496.47,
                    "word": " and",
                    "probability": 0.8955078125
                },
                {
                    "start": 496.47,
                    "end": 496.95,
                    "word": " techniques",
                    "probability": 0.86083984375
                },
                {
                    "start": 496.95,
                    "end": 497.29,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 497.29,
                    "end": 497.79,
                    "word": " preserve",
                    "probability": 0.77685546875
                },
                {
                    "start": 497.79,
                    "end": 498.39,
                    "word": " privacy",
                    "probability": 0.83251953125
                },
                {
                    "start": 498.39,
                    "end": 498.69,
                    "word": ".",
                    "probability": 0.9833984375
                }
            ]
        },
        {
            "id": 80,
            "seek": 52280,
            "start": 499.65,
            "end": 507.89,
            "text": " To preserve privacy, we have four main points to protect. The training data and input and output",
            "tokens": [
                50412,
                1407,
                15665,
                11427,
                11,
                321,
                362,
                1451,
                2135,
                2793,
                281,
                2371,
                13,
                440,
                3097,
                1412,
                293,
                4846,
                293,
                5598,
                50816
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23926957831325302,
            "compression_ratio": 1.6754385964912282,
            "no_speech_prob": 0.0004456043243408203,
            "words": [
                {
                    "start": 499.65,
                    "end": 500.13,
                    "word": " To",
                    "probability": 0.74609375
                },
                {
                    "start": 500.13,
                    "end": 500.45,
                    "word": " preserve",
                    "probability": 0.59228515625
                },
                {
                    "start": 500.45,
                    "end": 501.03,
                    "word": " privacy",
                    "probability": 0.830078125
                },
                {
                    "start": 501.03,
                    "end": 501.51,
                    "word": ",",
                    "probability": 0.7392578125
                },
                {
                    "start": 501.51,
                    "end": 501.65,
                    "word": " we",
                    "probability": 0.90869140625
                },
                {
                    "start": 501.65,
                    "end": 502.07,
                    "word": " have",
                    "probability": 0.88134765625
                },
                {
                    "start": 502.07,
                    "end": 502.41,
                    "word": " four",
                    "probability": 0.814453125
                },
                {
                    "start": 502.41,
                    "end": 502.65,
                    "word": " main",
                    "probability": 0.80078125
                },
                {
                    "start": 502.65,
                    "end": 503.09,
                    "word": " points",
                    "probability": 0.77294921875
                },
                {
                    "start": 503.09,
                    "end": 503.33,
                    "word": " to",
                    "probability": 0.89892578125
                },
                {
                    "start": 503.33,
                    "end": 503.85,
                    "word": " protect",
                    "probability": 0.798828125
                },
                {
                    "start": 503.85,
                    "end": 504.05,
                    "word": ".",
                    "probability": 0.39892578125
                },
                {
                    "start": 504.05,
                    "end": 504.15,
                    "word": " The",
                    "probability": 0.78564453125
                },
                {
                    "start": 504.15,
                    "end": 504.51,
                    "word": " training",
                    "probability": 0.83447265625
                },
                {
                    "start": 504.51,
                    "end": 505.09,
                    "word": " data",
                    "probability": 0.81103515625
                },
                {
                    "start": 505.09,
                    "end": 506.15,
                    "word": " and",
                    "probability": 0.509765625
                },
                {
                    "start": 506.15,
                    "end": 506.57,
                    "word": " input",
                    "probability": 0.8076171875
                },
                {
                    "start": 506.57,
                    "end": 507.55,
                    "word": " and",
                    "probability": 0.60791015625
                },
                {
                    "start": 507.55,
                    "end": 507.89,
                    "word": " output",
                    "probability": 0.7958984375
                }
            ]
        },
        {
            "id": 81,
            "seek": 52280,
            "start": 507.89,
            "end": 513.91,
            "text": " of the model shouldn't be visible to anyone who is not authorized to visit them. And finally,",
            "tokens": [
                50816,
                295,
                264,
                2316,
                4659,
                380,
                312,
                8974,
                281,
                2878,
                567,
                307,
                406,
                28312,
                281,
                3441,
                552,
                13,
                400,
                2721,
                11,
                51112
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23926957831325302,
            "compression_ratio": 1.6754385964912282,
            "no_speech_prob": 0.0004456043243408203,
            "words": [
                {
                    "start": 507.89,
                    "end": 508.09,
                    "word": " of",
                    "probability": 0.8984375
                },
                {
                    "start": 508.09,
                    "end": 508.25,
                    "word": " the",
                    "probability": 0.8154296875
                },
                {
                    "start": 508.25,
                    "end": 508.57,
                    "word": " model",
                    "probability": 0.75732421875
                },
                {
                    "start": 508.57,
                    "end": 509.05,
                    "word": " shouldn't",
                    "probability": 0.90283203125
                },
                {
                    "start": 509.05,
                    "end": 509.27,
                    "word": " be",
                    "probability": 0.90625
                },
                {
                    "start": 509.27,
                    "end": 509.83,
                    "word": " visible",
                    "probability": 0.88623046875
                },
                {
                    "start": 509.83,
                    "end": 510.31,
                    "word": " to",
                    "probability": 0.896484375
                },
                {
                    "start": 510.31,
                    "end": 510.69,
                    "word": " anyone",
                    "probability": 0.90380859375
                },
                {
                    "start": 510.69,
                    "end": 510.95,
                    "word": " who",
                    "probability": 0.80615234375
                },
                {
                    "start": 510.95,
                    "end": 511.07,
                    "word": " is",
                    "probability": 0.88330078125
                },
                {
                    "start": 511.07,
                    "end": 511.27,
                    "word": " not",
                    "probability": 0.88916015625
                },
                {
                    "start": 511.27,
                    "end": 511.65,
                    "word": " authorized",
                    "probability": 0.7275390625
                },
                {
                    "start": 511.65,
                    "end": 511.93,
                    "word": " to",
                    "probability": 0.904296875
                },
                {
                    "start": 511.93,
                    "end": 512.15,
                    "word": " visit",
                    "probability": 0.84716796875
                },
                {
                    "start": 512.15,
                    "end": 512.53,
                    "word": " them",
                    "probability": 0.73779296875
                },
                {
                    "start": 512.53,
                    "end": 513.03,
                    "word": ".",
                    "probability": 0.77880859375
                },
                {
                    "start": 513.03,
                    "end": 513.41,
                    "word": " And",
                    "probability": 0.8251953125
                },
                {
                    "start": 513.41,
                    "end": 513.81,
                    "word": " finally",
                    "probability": 0.73486328125
                },
                {
                    "start": 513.81,
                    "end": 513.91,
                    "word": ",",
                    "probability": 0.75732421875
                }
            ]
        },
        {
            "id": 82,
            "seek": 52280,
            "start": 513.91,
            "end": 519.93,
            "text": " the model privacy, which makes sure that the model cannot be stolen by any malicious party.",
            "tokens": [
                51112,
                264,
                2316,
                11427,
                11,
                597,
                1669,
                988,
                300,
                264,
                2316,
                2644,
                312,
                15900,
                538,
                604,
                33496,
                3595,
                13,
                51384
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23926957831325302,
            "compression_ratio": 1.6754385964912282,
            "no_speech_prob": 0.0004456043243408203,
            "words": [
                {
                    "start": 513.91,
                    "end": 514.05,
                    "word": " the",
                    "probability": 0.8134765625
                },
                {
                    "start": 514.05,
                    "end": 514.25,
                    "word": " model",
                    "probability": 0.8349609375
                },
                {
                    "start": 514.25,
                    "end": 514.85,
                    "word": " privacy",
                    "probability": 0.8154296875
                },
                {
                    "start": 514.85,
                    "end": 515.07,
                    "word": ",",
                    "probability": 0.6787109375
                },
                {
                    "start": 515.07,
                    "end": 515.31,
                    "word": " which",
                    "probability": 0.91259765625
                },
                {
                    "start": 515.31,
                    "end": 515.75,
                    "word": " makes",
                    "probability": 0.6982421875
                },
                {
                    "start": 515.75,
                    "end": 516.13,
                    "word": " sure",
                    "probability": 0.81787109375
                },
                {
                    "start": 516.13,
                    "end": 516.35,
                    "word": " that",
                    "probability": 0.87841796875
                },
                {
                    "start": 516.35,
                    "end": 516.53,
                    "word": " the",
                    "probability": 0.814453125
                },
                {
                    "start": 516.53,
                    "end": 516.79,
                    "word": " model",
                    "probability": 0.81298828125
                },
                {
                    "start": 516.79,
                    "end": 517.13,
                    "word": " cannot",
                    "probability": 0.76953125
                },
                {
                    "start": 517.13,
                    "end": 517.39,
                    "word": " be",
                    "probability": 0.90380859375
                },
                {
                    "start": 517.39,
                    "end": 517.75,
                    "word": " stolen",
                    "probability": 0.7509765625
                },
                {
                    "start": 517.75,
                    "end": 518.09,
                    "word": " by",
                    "probability": 0.90380859375
                },
                {
                    "start": 518.09,
                    "end": 518.39,
                    "word": " any",
                    "probability": 0.78466796875
                },
                {
                    "start": 518.39,
                    "end": 518.69,
                    "word": " malicious",
                    "probability": 0.82080078125
                },
                {
                    "start": 518.69,
                    "end": 519.21,
                    "word": " party",
                    "probability": 0.71923828125
                },
                {
                    "start": 519.21,
                    "end": 519.93,
                    "word": ".",
                    "probability": 0.9921875
                }
            ]
        },
        {
            "id": 83,
            "seek": 52280,
            "start": 520.17,
            "end": 527.11,
            "text": " Two main categories to preserve privacy are secure computation and privacy preservation. In case of",
            "tokens": [
                51424,
                4453,
                2135,
                10479,
                281,
                15665,
                11427,
                366,
                7144,
                24903,
                293,
                11427,
                27257,
                13,
                682,
                1389,
                295,
                51772
            ],
            "temperature": 0.0,
            "avg_logprob": -0.23926957831325302,
            "compression_ratio": 1.6754385964912282,
            "no_speech_prob": 0.0004456043243408203,
            "words": [
                {
                    "start": 520.17,
                    "end": 520.39,
                    "word": " Two",
                    "probability": 0.8349609375
                },
                {
                    "start": 520.39,
                    "end": 520.85,
                    "word": " main",
                    "probability": 0.8427734375
                },
                {
                    "start": 520.85,
                    "end": 521.35,
                    "word": " categories",
                    "probability": 0.76318359375
                },
                {
                    "start": 521.35,
                    "end": 521.95,
                    "word": " to",
                    "probability": 0.9052734375
                },
                {
                    "start": 521.95,
                    "end": 522.47,
                    "word": " preserve",
                    "probability": 0.74462890625
                },
                {
                    "start": 522.47,
                    "end": 523.05,
                    "word": " privacy",
                    "probability": 0.8310546875
                },
                {
                    "start": 523.05,
                    "end": 523.51,
                    "word": " are",
                    "probability": 0.869140625
                },
                {
                    "start": 523.51,
                    "end": 523.89,
                    "word": " secure",
                    "probability": 0.79443359375
                },
                {
                    "start": 523.89,
                    "end": 524.39,
                    "word": " computation",
                    "probability": 0.8125
                },
                {
                    "start": 524.39,
                    "end": 525.19,
                    "word": " and",
                    "probability": 0.8798828125
                },
                {
                    "start": 525.19,
                    "end": 525.61,
                    "word": " privacy",
                    "probability": 0.83544921875
                },
                {
                    "start": 525.61,
                    "end": 526.31,
                    "word": " preservation",
                    "probability": 0.73828125
                },
                {
                    "start": 526.31,
                    "end": 526.57,
                    "word": ".",
                    "probability": 0.97998046875
                },
                {
                    "start": 526.57,
                    "end": 526.75,
                    "word": " In",
                    "probability": 0.88818359375
                },
                {
                    "start": 526.75,
                    "end": 526.95,
                    "word": " case",
                    "probability": 0.7607421875
                },
                {
                    "start": 526.95,
                    "end": 527.11,
                    "word": " of",
                    "probability": 0.904296875
                }
            ]
        },
        {
            "id": 84,
            "seek": 54760,
            "start": 527.11,
            "end": 534.09,
            "text": " secure computation, we transform or distribute our computations in a way that is not readable",
            "tokens": [
                50364,
                7144,
                24903,
                11,
                321,
                4088,
                420,
                20594,
                527,
                2807,
                763,
                294,
                257,
                636,
                300,
                307,
                406,
                49857,
                50716
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17783717791500844,
            "compression_ratio": 1.7772277227722773,
            "no_speech_prob": 0.0014743804931640625,
            "words": [
                {
                    "start": 527.11,
                    "end": 527.43,
                    "word": " secure",
                    "probability": 0.343994140625
                },
                {
                    "start": 527.43,
                    "end": 528.17,
                    "word": " computation",
                    "probability": 0.7783203125
                },
                {
                    "start": 528.17,
                    "end": 528.63,
                    "word": ",",
                    "probability": 0.7109375
                },
                {
                    "start": 528.63,
                    "end": 529.23,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 529.23,
                    "end": 529.95,
                    "word": " transform",
                    "probability": 0.8876953125
                },
                {
                    "start": 529.95,
                    "end": 530.61,
                    "word": " or",
                    "probability": 0.8828125
                },
                {
                    "start": 530.61,
                    "end": 531.21,
                    "word": " distribute",
                    "probability": 0.7734375
                },
                {
                    "start": 531.21,
                    "end": 531.65,
                    "word": " our",
                    "probability": 0.783203125
                },
                {
                    "start": 531.65,
                    "end": 532.53,
                    "word": " computations",
                    "probability": 0.907958984375
                },
                {
                    "start": 532.53,
                    "end": 532.81,
                    "word": " in",
                    "probability": 0.89990234375
                },
                {
                    "start": 532.97,
                    "end": 533.15,
                    "word": " a way",
                    "probability": 0.88134765625
                },
                {
                    "start": 533.15,
                    "end": 533.37,
                    "word": " that",
                    "probability": 0.87939453125
                },
                {
                    "start": 533.37,
                    "end": 533.55,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 533.55,
                    "end": 533.77,
                    "word": " not",
                    "probability": 0.89697265625
                },
                {
                    "start": 533.77,
                    "end": 534.09,
                    "word": " readable",
                    "probability": 0.83642578125
                }
            ]
        },
        {
            "id": 85,
            "seek": 54760,
            "start": 534.09,
            "end": 540.33,
            "text": " by any unauthorized person. And in terms of privacy preservation techniques, we use injection,",
            "tokens": [
                50716,
                538,
                604,
                517,
                34224,
                1602,
                954,
                13,
                400,
                294,
                2115,
                295,
                11427,
                27257,
                7512,
                11,
                321,
                764,
                22873,
                11,
                51024
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17783717791500844,
            "compression_ratio": 1.7772277227722773,
            "no_speech_prob": 0.0014743804931640625,
            "words": [
                {
                    "start": 534.09,
                    "end": 534.43,
                    "word": " by",
                    "probability": 0.9072265625
                },
                {
                    "start": 534.43,
                    "end": 534.69,
                    "word": " any",
                    "probability": 0.82373046875
                },
                {
                    "start": 534.69,
                    "end": 535.33,
                    "word": " unauthorized",
                    "probability": 0.8728841145833334
                },
                {
                    "start": 535.33,
                    "end": 535.83,
                    "word": " person",
                    "probability": 0.783203125
                },
                {
                    "start": 535.83,
                    "end": 536.27,
                    "word": ".",
                    "probability": 0.765625
                },
                {
                    "start": 536.27,
                    "end": 536.35,
                    "word": " And",
                    "probability": 0.845703125
                },
                {
                    "start": 536.35,
                    "end": 536.55,
                    "word": " in",
                    "probability": 0.89697265625
                },
                {
                    "start": 536.55,
                    "end": 536.79,
                    "word": " terms",
                    "probability": 0.77294921875
                },
                {
                    "start": 536.79,
                    "end": 536.95,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 536.95,
                    "end": 537.33,
                    "word": " privacy",
                    "probability": 0.8056640625
                },
                {
                    "start": 537.33,
                    "end": 537.87,
                    "word": " preservation",
                    "probability": 0.65478515625
                },
                {
                    "start": 537.87,
                    "end": 538.65,
                    "word": " techniques",
                    "probability": 0.84130859375
                },
                {
                    "start": 538.65,
                    "end": 538.79,
                    "word": ",",
                    "probability": 0.9306640625
                },
                {
                    "start": 538.79,
                    "end": 538.99,
                    "word": " we",
                    "probability": 0.89208984375
                },
                {
                    "start": 538.99,
                    "end": 539.53,
                    "word": " use",
                    "probability": 0.82861328125
                },
                {
                    "start": 539.53,
                    "end": 540.11,
                    "word": " injection",
                    "probability": 0.64404296875
                },
                {
                    "start": 540.11,
                    "end": 540.33,
                    "word": ",",
                    "probability": 0.75390625
                }
            ]
        },
        {
            "id": 86,
            "seek": 54760,
            "start": 540.33,
            "end": 544.95,
            "text": " noise injection, or masking and hiding techniques in order to preserve privacy.",
            "tokens": [
                51024,
                5658,
                22873,
                11,
                420,
                31226,
                293,
                10596,
                7512,
                294,
                1668,
                281,
                15665,
                11427,
                13,
                51240
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17783717791500844,
            "compression_ratio": 1.7772277227722773,
            "no_speech_prob": 0.0014743804931640625,
            "words": [
                {
                    "start": 540.33,
                    "end": 540.67,
                    "word": " noise",
                    "probability": 0.88330078125
                },
                {
                    "start": 540.67,
                    "end": 541.11,
                    "word": " injection",
                    "probability": 0.6845703125
                },
                {
                    "start": 541.11,
                    "end": 541.33,
                    "word": ",",
                    "probability": 0.5615234375
                },
                {
                    "start": 541.33,
                    "end": 541.51,
                    "word": " or",
                    "probability": 0.90478515625
                },
                {
                    "start": 541.51,
                    "end": 541.91,
                    "word": " masking",
                    "probability": 0.724609375
                },
                {
                    "start": 541.91,
                    "end": 542.19,
                    "word": " and",
                    "probability": 0.880859375
                },
                {
                    "start": 542.19,
                    "end": 542.45,
                    "word": " hiding",
                    "probability": 0.7392578125
                },
                {
                    "start": 542.45,
                    "end": 542.97,
                    "word": " techniques",
                    "probability": 0.82958984375
                },
                {
                    "start": 542.97,
                    "end": 543.21,
                    "word": " in",
                    "probability": 0.8984375
                },
                {
                    "start": 543.21,
                    "end": 543.39,
                    "word": " order",
                    "probability": 0.8212890625
                },
                {
                    "start": 543.39,
                    "end": 543.65,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 543.65,
                    "end": 544.01,
                    "word": " preserve",
                    "probability": 0.775390625
                },
                {
                    "start": 544.01,
                    "end": 544.59,
                    "word": " privacy",
                    "probability": 0.84814453125
                },
                {
                    "start": 544.59,
                    "end": 544.95,
                    "word": ".",
                    "probability": 0.99267578125
                }
            ]
        },
        {
            "id": 87,
            "seek": 54760,
            "start": 545.79,
            "end": 551.91,
            "text": " The first method of privacy preservation is homomorphic encryption. Homomorphic encryption",
            "tokens": [
                51292,
                440,
                700,
                3170,
                295,
                11427,
                27257,
                307,
                3655,
                32702,
                299,
                29575,
                13,
                20903,
                32702,
                299,
                29575,
                51612
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17783717791500844,
            "compression_ratio": 1.7772277227722773,
            "no_speech_prob": 0.0014743804931640625,
            "words": [
                {
                    "start": 545.79,
                    "end": 545.97,
                    "word": " The",
                    "probability": 0.76171875
                },
                {
                    "start": 545.97,
                    "end": 546.27,
                    "word": " first",
                    "probability": 0.8125
                },
                {
                    "start": 546.27,
                    "end": 546.61,
                    "word": " method",
                    "probability": 0.82421875
                },
                {
                    "start": 546.61,
                    "end": 547.27,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 547.27,
                    "end": 548.77,
                    "word": " privacy",
                    "probability": 0.83203125
                },
                {
                    "start": 548.77,
                    "end": 549.25,
                    "word": " preservation",
                    "probability": 0.7060546875
                },
                {
                    "start": 549.25,
                    "end": 549.77,
                    "word": " is",
                    "probability": 0.8955078125
                },
                {
                    "start": 549.77,
                    "end": 550.35,
                    "word": " homomorphic",
                    "probability": 0.9342447916666666
                },
                {
                    "start": 550.35,
                    "end": 550.77,
                    "word": " encryption",
                    "probability": 0.76123046875
                },
                {
                    "start": 550.77,
                    "end": 551.13,
                    "word": ".",
                    "probability": 0.93212890625
                },
                {
                    "start": 551.13,
                    "end": 551.57,
                    "word": " Homomorphic",
                    "probability": 0.9423828125
                },
                {
                    "start": 551.57,
                    "end": 551.91,
                    "word": " encryption",
                    "probability": 0.7626953125
                }
            ]
        },
        {
            "id": 88,
            "seek": 57670,
            "start": 551.91,
            "end": 558.61,
            "text": " is a type of encryption where you are doing the same computations and a third party can",
            "tokens": [
                50372,
                307,
                257,
                2010,
                295,
                29575,
                689,
                291,
                366,
                884,
                264,
                912,
                2807,
                763,
                293,
                257,
                2636,
                3595,
                393,
                50696
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2156593462268075,
            "compression_ratio": 1.8067632850241546,
            "no_speech_prob": 0.007770538330078125,
            "words": [
                {
                    "start": 551.91,
                    "end": 552.23,
                    "word": " is",
                    "probability": 0.08526611328125
                },
                {
                    "start": 552.37,
                    "end": 552.61,
                    "word": " a type",
                    "probability": 0.90087890625
                },
                {
                    "start": 552.61,
                    "end": 552.77,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 552.77,
                    "end": 553.19,
                    "word": " encryption",
                    "probability": 0.7880859375
                },
                {
                    "start": 553.19,
                    "end": 554.27,
                    "word": " where",
                    "probability": 0.78466796875
                },
                {
                    "start": 554.27,
                    "end": 554.93,
                    "word": " you",
                    "probability": 0.8896484375
                },
                {
                    "start": 554.93,
                    "end": 555.19,
                    "word": " are",
                    "probability": 0.890625
                },
                {
                    "start": 555.19,
                    "end": 555.45,
                    "word": " doing",
                    "probability": 0.8916015625
                },
                {
                    "start": 555.45,
                    "end": 555.65,
                    "word": " the",
                    "probability": 0.8173828125
                },
                {
                    "start": 555.65,
                    "end": 555.89,
                    "word": " same",
                    "probability": 0.81494140625
                },
                {
                    "start": 555.89,
                    "end": 556.71,
                    "word": " computations",
                    "probability": 0.91845703125
                },
                {
                    "start": 556.71,
                    "end": 557.65,
                    "word": " and",
                    "probability": 0.73974609375
                },
                {
                    "start": 557.89,
                    "end": 558.07,
                    "word": " a third",
                    "probability": 0.8125
                },
                {
                    "start": 558.07,
                    "end": 558.35,
                    "word": " party",
                    "probability": 0.7177734375
                },
                {
                    "start": 558.35,
                    "end": 558.61,
                    "word": " can",
                    "probability": 0.91015625
                }
            ]
        },
        {
            "id": 89,
            "seek": 57670,
            "start": 558.61,
            "end": 564.13,
            "text": " do your computations for you, but not on the original form of the data, on an encrypted form",
            "tokens": [
                50696,
                360,
                428,
                2807,
                763,
                337,
                291,
                11,
                457,
                406,
                322,
                264,
                3380,
                1254,
                295,
                264,
                1412,
                11,
                322,
                364,
                36663,
                1254,
                50972
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2156593462268075,
            "compression_ratio": 1.8067632850241546,
            "no_speech_prob": 0.007770538330078125,
            "words": [
                {
                    "start": 558.61,
                    "end": 558.79,
                    "word": " do",
                    "probability": 0.9091796875
                },
                {
                    "start": 558.79,
                    "end": 558.97,
                    "word": " your",
                    "probability": 0.78369140625
                },
                {
                    "start": 558.97,
                    "end": 559.63,
                    "word": " computations",
                    "probability": 0.906494140625
                },
                {
                    "start": 559.63,
                    "end": 559.99,
                    "word": " for",
                    "probability": 0.89892578125
                },
                {
                    "start": 559.99,
                    "end": 560.31,
                    "word": " you",
                    "probability": 0.896484375
                },
                {
                    "start": 560.31,
                    "end": 560.57,
                    "word": ",",
                    "probability": 0.4716796875
                },
                {
                    "start": 560.57,
                    "end": 560.77,
                    "word": " but",
                    "probability": 0.89404296875
                },
                {
                    "start": 560.77,
                    "end": 561.09,
                    "word": " not",
                    "probability": 0.8896484375
                },
                {
                    "start": 561.09,
                    "end": 561.31,
                    "word": " on",
                    "probability": 0.89208984375
                },
                {
                    "start": 561.31,
                    "end": 561.43,
                    "word": " the",
                    "probability": 0.81005859375
                },
                {
                    "start": 561.43,
                    "end": 561.75,
                    "word": " original",
                    "probability": 0.7607421875
                },
                {
                    "start": 561.75,
                    "end": 562.17,
                    "word": " form",
                    "probability": 0.8486328125
                },
                {
                    "start": 562.17,
                    "end": 562.37,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 562.37,
                    "end": 562.49,
                    "word": " the",
                    "probability": 0.80322265625
                },
                {
                    "start": 562.49,
                    "end": 562.81,
                    "word": " data",
                    "probability": 0.8076171875
                },
                {
                    "start": 562.81,
                    "end": 563.09,
                    "word": ",",
                    "probability": 0.40087890625
                },
                {
                    "start": 563.09,
                    "end": 563.19,
                    "word": " on",
                    "probability": 0.88525390625
                },
                {
                    "start": 563.19,
                    "end": 563.37,
                    "word": " an",
                    "probability": 0.8935546875
                },
                {
                    "start": 563.37,
                    "end": 563.73,
                    "word": " encrypted",
                    "probability": 0.900390625
                },
                {
                    "start": 563.73,
                    "end": 564.13,
                    "word": " form",
                    "probability": 0.82421875
                }
            ]
        },
        {
            "id": 90,
            "seek": 57670,
            "start": 564.13,
            "end": 575.29,
            "text": " of data. This means that if you use a third party company for your ML computations, you can totally",
            "tokens": [
                50972,
                295,
                1412,
                13,
                639,
                1355,
                300,
                498,
                291,
                764,
                257,
                2636,
                3595,
                2237,
                337,
                428,
                21601,
                2807,
                763,
                11,
                291,
                393,
                3879,
                51536
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2156593462268075,
            "compression_ratio": 1.8067632850241546,
            "no_speech_prob": 0.007770538330078125,
            "words": [
                {
                    "start": 564.13,
                    "end": 564.33,
                    "word": " of",
                    "probability": 0.900390625
                },
                {
                    "start": 564.33,
                    "end": 564.67,
                    "word": " data",
                    "probability": 0.350830078125
                },
                {
                    "start": 564.67,
                    "end": 565.13,
                    "word": ".",
                    "probability": 0.97314453125
                },
                {
                    "start": 565.13,
                    "end": 565.51,
                    "word": " This",
                    "probability": 0.80419921875
                },
                {
                    "start": 565.51,
                    "end": 565.79,
                    "word": " means",
                    "probability": 0.8798828125
                },
                {
                    "start": 565.79,
                    "end": 566.03,
                    "word": " that",
                    "probability": 0.87353515625
                },
                {
                    "start": 566.03,
                    "end": 566.31,
                    "word": " if",
                    "probability": 0.88525390625
                },
                {
                    "start": 566.31,
                    "end": 566.53,
                    "word": " you",
                    "probability": 0.89013671875
                },
                {
                    "start": 566.53,
                    "end": 568.73,
                    "word": " use",
                    "probability": 0.59130859375
                },
                {
                    "start": 568.89,
                    "end": 569.11,
                    "word": " a third",
                    "probability": 0.8369140625
                },
                {
                    "start": 569.11,
                    "end": 569.53,
                    "word": " party",
                    "probability": 0.5693359375
                },
                {
                    "start": 569.53,
                    "end": 570.47,
                    "word": " company",
                    "probability": 0.7919921875
                },
                {
                    "start": 570.47,
                    "end": 571.19,
                    "word": " for",
                    "probability": 0.896484375
                },
                {
                    "start": 571.19,
                    "end": 572.11,
                    "word": " your",
                    "probability": 0.78662109375
                },
                {
                    "start": 572.11,
                    "end": 573.09,
                    "word": " ML",
                    "probability": 0.548828125
                },
                {
                    "start": 573.09,
                    "end": 574.01,
                    "word": " computations",
                    "probability": 0.90869140625
                },
                {
                    "start": 574.01,
                    "end": 574.31,
                    "word": ",",
                    "probability": 0.86328125
                },
                {
                    "start": 574.31,
                    "end": 574.59,
                    "word": " you",
                    "probability": 0.89501953125
                },
                {
                    "start": 574.59,
                    "end": 574.85,
                    "word": " can",
                    "probability": 0.88623046875
                },
                {
                    "start": 574.85,
                    "end": 575.29,
                    "word": " totally",
                    "probability": 0.70263671875
                }
            ]
        },
        {
            "id": 91,
            "seek": 57670,
            "start": 575.29,
            "end": 581.01,
            "text": " give your data to them in a transformed form, which is using homomorphic encryption, and they",
            "tokens": [
                51536,
                976,
                428,
                1412,
                281,
                552,
                294,
                257,
                16894,
                1254,
                11,
                597,
                307,
                1228,
                3655,
                32702,
                299,
                29575,
                11,
                293,
                436,
                51812
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2156593462268075,
            "compression_ratio": 1.8067632850241546,
            "no_speech_prob": 0.007770538330078125,
            "words": [
                {
                    "start": 575.29,
                    "end": 575.67,
                    "word": " give",
                    "probability": 0.8212890625
                },
                {
                    "start": 575.67,
                    "end": 575.83,
                    "word": " your",
                    "probability": 0.7919921875
                },
                {
                    "start": 575.83,
                    "end": 576.11,
                    "word": " data",
                    "probability": 0.802734375
                },
                {
                    "start": 576.11,
                    "end": 576.33,
                    "word": " to",
                    "probability": 0.9052734375
                },
                {
                    "start": 576.33,
                    "end": 576.55,
                    "word": " them",
                    "probability": 0.73095703125
                },
                {
                    "start": 576.55,
                    "end": 576.81,
                    "word": " in",
                    "probability": 0.90283203125
                },
                {
                    "start": 576.95,
                    "end": 577.61,
                    "word": " a transformed",
                    "probability": 0.192138671875
                },
                {
                    "start": 577.61,
                    "end": 578.21,
                    "word": " form",
                    "probability": 0.814453125
                },
                {
                    "start": 578.21,
                    "end": 578.41,
                    "word": ",",
                    "probability": 0.7548828125
                },
                {
                    "start": 578.41,
                    "end": 578.79,
                    "word": " which",
                    "probability": 0.91552734375
                },
                {
                    "start": 578.79,
                    "end": 578.95,
                    "word": " is",
                    "probability": 0.9052734375
                },
                {
                    "start": 578.95,
                    "end": 579.37,
                    "word": " using",
                    "probability": 0.8310546875
                },
                {
                    "start": 579.37,
                    "end": 579.97,
                    "word": " homomorphic",
                    "probability": 0.9181315104166666
                },
                {
                    "start": 579.97,
                    "end": 580.41,
                    "word": " encryption",
                    "probability": 0.81005859375
                },
                {
                    "start": 580.41,
                    "end": 580.41,
                    "word": ",",
                    "probability": 0.322509765625
                },
                {
                    "start": 580.41,
                    "end": 580.85,
                    "word": " and",
                    "probability": 0.8916015625
                },
                {
                    "start": 580.85,
                    "end": 581.01,
                    "word": " they",
                    "probability": 0.73876953125
                }
            ]
        },
        {
            "id": 92,
            "seek": 60324,
            "start": 581.01,
            "end": 585.79,
            "text": " can do the computations for you and give you back the results. And then you can decrypt the data",
            "tokens": [
                50364,
                393,
                360,
                264,
                2807,
                763,
                337,
                291,
                293,
                976,
                291,
                646,
                264,
                3542,
                13,
                400,
                550,
                291,
                393,
                979,
                627,
                662,
                264,
                1412,
                50612
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19220430684346024,
            "compression_ratio": 1.6637554585152838,
            "no_speech_prob": 0.00208282470703125,
            "words": [
                {
                    "start": 581.01,
                    "end": 581.17,
                    "word": " can",
                    "probability": 0.1070556640625
                },
                {
                    "start": 581.17,
                    "end": 581.29,
                    "word": " do",
                    "probability": 0.461181640625
                },
                {
                    "start": 581.29,
                    "end": 581.45,
                    "word": " the",
                    "probability": 0.76220703125
                },
                {
                    "start": 581.45,
                    "end": 582.13,
                    "word": " computations",
                    "probability": 0.8828125
                },
                {
                    "start": 582.13,
                    "end": 582.43,
                    "word": " for",
                    "probability": 0.875
                },
                {
                    "start": 582.43,
                    "end": 582.63,
                    "word": " you",
                    "probability": 0.892578125
                },
                {
                    "start": 582.63,
                    "end": 582.85,
                    "word": " and",
                    "probability": 0.84375
                },
                {
                    "start": 582.85,
                    "end": 583.07,
                    "word": " give",
                    "probability": 0.78076171875
                },
                {
                    "start": 583.07,
                    "end": 583.21,
                    "word": " you",
                    "probability": 0.884765625
                },
                {
                    "start": 583.21,
                    "end": 583.47,
                    "word": " back",
                    "probability": 0.7890625
                },
                {
                    "start": 583.47,
                    "end": 583.63,
                    "word": " the",
                    "probability": 0.80126953125
                },
                {
                    "start": 583.63,
                    "end": 584.03,
                    "word": " results",
                    "probability": 0.689453125
                },
                {
                    "start": 584.03,
                    "end": 584.27,
                    "word": ".",
                    "probability": 0.51953125
                },
                {
                    "start": 584.27,
                    "end": 584.37,
                    "word": " And",
                    "probability": 0.767578125
                },
                {
                    "start": 584.37,
                    "end": 584.59,
                    "word": " then",
                    "probability": 0.72802734375
                },
                {
                    "start": 584.59,
                    "end": 584.77,
                    "word": " you",
                    "probability": 0.84814453125
                },
                {
                    "start": 584.77,
                    "end": 585.01,
                    "word": " can",
                    "probability": 0.8955078125
                },
                {
                    "start": 585.01,
                    "end": 585.37,
                    "word": " decrypt",
                    "probability": 0.8873697916666666
                },
                {
                    "start": 585.37,
                    "end": 585.51,
                    "word": " the",
                    "probability": 0.79931640625
                },
                {
                    "start": 585.51,
                    "end": 585.79,
                    "word": " data",
                    "probability": 0.8154296875
                }
            ]
        },
        {
            "id": 93,
            "seek": 60324,
            "start": 585.79,
            "end": 592.47,
            "text": " and get the plain text or cipher text as you want. The second method is secure multi-party",
            "tokens": [
                50612,
                293,
                483,
                264,
                11121,
                2487,
                420,
                269,
                21240,
                2487,
                382,
                291,
                528,
                13,
                440,
                1150,
                3170,
                307,
                7144,
                4825,
                12,
                23409,
                50940
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19220430684346024,
            "compression_ratio": 1.6637554585152838,
            "no_speech_prob": 0.00208282470703125,
            "words": [
                {
                    "start": 585.79,
                    "end": 586.15,
                    "word": " and",
                    "probability": 0.86865234375
                },
                {
                    "start": 586.15,
                    "end": 586.33,
                    "word": " get",
                    "probability": 0.89111328125
                },
                {
                    "start": 586.33,
                    "end": 586.51,
                    "word": " the",
                    "probability": 0.7158203125
                },
                {
                    "start": 586.51,
                    "end": 586.77,
                    "word": " plain",
                    "probability": 0.91748046875
                },
                {
                    "start": 586.77,
                    "end": 587.09,
                    "word": " text",
                    "probability": 0.4990234375
                },
                {
                    "start": 587.09,
                    "end": 587.35,
                    "word": " or",
                    "probability": 0.890625
                },
                {
                    "start": 587.35,
                    "end": 587.61,
                    "word": " cipher",
                    "probability": 0.82861328125
                },
                {
                    "start": 587.61,
                    "end": 587.87,
                    "word": " text",
                    "probability": 0.28955078125
                },
                {
                    "start": 587.87,
                    "end": 588.25,
                    "word": " as",
                    "probability": 0.876953125
                },
                {
                    "start": 588.25,
                    "end": 588.45,
                    "word": " you",
                    "probability": 0.8916015625
                },
                {
                    "start": 588.45,
                    "end": 588.77,
                    "word": " want",
                    "probability": 0.8193359375
                },
                {
                    "start": 588.77,
                    "end": 589.35,
                    "word": ".",
                    "probability": 0.9892578125
                },
                {
                    "start": 590.33,
                    "end": 590.71,
                    "word": " The",
                    "probability": 0.802734375
                },
                {
                    "start": 590.71,
                    "end": 590.95,
                    "word": " second",
                    "probability": 0.76025390625
                },
                {
                    "start": 590.95,
                    "end": 591.23,
                    "word": " method",
                    "probability": 0.8291015625
                },
                {
                    "start": 591.23,
                    "end": 591.51,
                    "word": " is",
                    "probability": 0.90087890625
                },
                {
                    "start": 591.51,
                    "end": 591.85,
                    "word": " secure",
                    "probability": 0.76611328125
                },
                {
                    "start": 591.85,
                    "end": 592.17,
                    "word": " multi",
                    "probability": 0.472900390625
                },
                {
                    "start": 592.17,
                    "end": 592.47,
                    "word": "-party",
                    "probability": 0.837646484375
                }
            ]
        },
        {
            "id": 94,
            "seek": 60324,
            "start": 592.47,
            "end": 599.73,
            "text": " computation. In case of this technique, we have multiple parties who all contribute into building",
            "tokens": [
                50940,
                24903,
                13,
                682,
                1389,
                295,
                341,
                6532,
                11,
                321,
                362,
                3866,
                8265,
                567,
                439,
                10586,
                666,
                2390,
                51300
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19220430684346024,
            "compression_ratio": 1.6637554585152838,
            "no_speech_prob": 0.00208282470703125,
            "words": [
                {
                    "start": 592.47,
                    "end": 593.17,
                    "word": " computation",
                    "probability": 0.82275390625
                },
                {
                    "start": 593.17,
                    "end": 593.45,
                    "word": ".",
                    "probability": 0.97314453125
                },
                {
                    "start": 593.45,
                    "end": 593.67,
                    "word": " In",
                    "probability": 0.89794921875
                },
                {
                    "start": 593.67,
                    "end": 593.93,
                    "word": " case",
                    "probability": 0.7568359375
                },
                {
                    "start": 593.93,
                    "end": 594.25,
                    "word": " of",
                    "probability": 0.896484375
                },
                {
                    "start": 594.25,
                    "end": 594.57,
                    "word": " this",
                    "probability": 0.85009765625
                },
                {
                    "start": 594.57,
                    "end": 595.09,
                    "word": " technique",
                    "probability": 0.9072265625
                },
                {
                    "start": 595.09,
                    "end": 595.53,
                    "word": ",",
                    "probability": 0.8984375
                },
                {
                    "start": 595.53,
                    "end": 595.65,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 595.65,
                    "end": 595.89,
                    "word": " have",
                    "probability": 0.87646484375
                },
                {
                    "start": 595.89,
                    "end": 596.29,
                    "word": " multiple",
                    "probability": 0.74755859375
                },
                {
                    "start": 596.29,
                    "end": 596.87,
                    "word": " parties",
                    "probability": 0.85498046875
                },
                {
                    "start": 596.87,
                    "end": 597.69,
                    "word": " who",
                    "probability": 0.796875
                },
                {
                    "start": 597.69,
                    "end": 598.13,
                    "word": " all",
                    "probability": 0.87890625
                },
                {
                    "start": 598.13,
                    "end": 598.63,
                    "word": " contribute",
                    "probability": 0.890625
                },
                {
                    "start": 598.63,
                    "end": 599.29,
                    "word": " into",
                    "probability": 0.6923828125
                },
                {
                    "start": 599.29,
                    "end": 599.73,
                    "word": " building",
                    "probability": 0.78076171875
                }
            ]
        },
        {
            "id": 95,
            "seek": 60324,
            "start": 599.73,
            "end": 607.55,
            "text": " an ML model or a function or to compute a value for a function. But the thing is that the input",
            "tokens": [
                51300,
                364,
                21601,
                2316,
                420,
                257,
                2445,
                420,
                281,
                14722,
                257,
                2158,
                337,
                257,
                2445,
                13,
                583,
                264,
                551,
                307,
                300,
                264,
                4846,
                51696
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19220430684346024,
            "compression_ratio": 1.6637554585152838,
            "no_speech_prob": 0.00208282470703125,
            "words": [
                {
                    "start": 599.73,
                    "end": 599.99,
                    "word": " an",
                    "probability": 0.8857421875
                },
                {
                    "start": 599.99,
                    "end": 600.21,
                    "word": " ML",
                    "probability": 0.8876953125
                },
                {
                    "start": 600.21,
                    "end": 600.59,
                    "word": " model",
                    "probability": 0.8408203125
                },
                {
                    "start": 600.59,
                    "end": 601.03,
                    "word": " or",
                    "probability": 0.62744140625
                },
                {
                    "start": 601.27,
                    "end": 601.69,
                    "word": " a function",
                    "probability": 0.88232421875
                },
                {
                    "start": 601.69,
                    "end": 602.19,
                    "word": " or",
                    "probability": 0.58984375
                },
                {
                    "start": 602.19,
                    "end": 602.39,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 602.39,
                    "end": 602.79,
                    "word": " compute",
                    "probability": 0.87109375
                },
                {
                    "start": 603.01,
                    "end": 603.35,
                    "word": " a value",
                    "probability": 0.93115234375
                },
                {
                    "start": 603.35,
                    "end": 603.77,
                    "word": " for",
                    "probability": 0.8984375
                },
                {
                    "start": 603.97,
                    "end": 604.31,
                    "word": " a function",
                    "probability": 0.88232421875
                },
                {
                    "start": 604.31,
                    "end": 604.81,
                    "word": ".",
                    "probability": 0.9423828125
                },
                {
                    "start": 604.81,
                    "end": 605.35,
                    "word": " But",
                    "probability": 0.880859375
                },
                {
                    "start": 605.35,
                    "end": 605.81,
                    "word": " the",
                    "probability": 0.76318359375
                },
                {
                    "start": 605.81,
                    "end": 606.03,
                    "word": " thing",
                    "probability": 0.81689453125
                },
                {
                    "start": 606.03,
                    "end": 606.27,
                    "word": " is",
                    "probability": 0.90966796875
                },
                {
                    "start": 606.27,
                    "end": 606.59,
                    "word": " that",
                    "probability": 0.869140625
                },
                {
                    "start": 606.59,
                    "end": 607.09,
                    "word": " the",
                    "probability": 0.74609375
                },
                {
                    "start": 607.09,
                    "end": 607.55,
                    "word": " input",
                    "probability": 0.8837890625
                }
            ]
        },
        {
            "id": 96,
            "seek": 63280,
            "start": 607.55,
            "end": 615.15,
            "text": " is distributed between all of these parties. So as you see, part of input A1 is here, A2 is here,",
            "tokens": [
                50368,
                307,
                12631,
                1296,
                439,
                295,
                613,
                8265,
                13,
                407,
                382,
                291,
                536,
                11,
                644,
                295,
                4846,
                316,
                16,
                307,
                510,
                11,
                316,
                17,
                307,
                510,
                11,
                50740
            ],
            "temperature": 0.0,
            "avg_logprob": -0.1833964583247599,
            "compression_ratio": 1.668103448275862,
            "no_speech_prob": 0.0025634765625,
            "words": [
                {
                    "start": 607.55,
                    "end": 607.85,
                    "word": " is",
                    "probability": 0.224365234375
                },
                {
                    "start": 607.85,
                    "end": 608.29,
                    "word": " distributed",
                    "probability": 0.689453125
                },
                {
                    "start": 608.29,
                    "end": 609.03,
                    "word": " between",
                    "probability": 0.74658203125
                },
                {
                    "start": 609.03,
                    "end": 609.37,
                    "word": " all",
                    "probability": 0.87841796875
                },
                {
                    "start": 609.37,
                    "end": 609.57,
                    "word": " of",
                    "probability": 0.87841796875
                },
                {
                    "start": 609.57,
                    "end": 609.79,
                    "word": " these",
                    "probability": 0.69873046875
                },
                {
                    "start": 609.79,
                    "end": 610.21,
                    "word": " parties",
                    "probability": 0.8701171875
                },
                {
                    "start": 610.21,
                    "end": 610.53,
                    "word": ".",
                    "probability": 0.7021484375
                },
                {
                    "start": 610.53,
                    "end": 610.55,
                    "word": " So",
                    "probability": 0.73291015625
                },
                {
                    "start": 610.55,
                    "end": 610.79,
                    "word": " as",
                    "probability": 0.70458984375
                },
                {
                    "start": 610.79,
                    "end": 610.95,
                    "word": " you",
                    "probability": 0.896484375
                },
                {
                    "start": 610.95,
                    "end": 611.31,
                    "word": " see",
                    "probability": 0.8017578125
                },
                {
                    "start": 611.31,
                    "end": 611.87,
                    "word": ",",
                    "probability": 0.5048828125
                },
                {
                    "start": 611.87,
                    "end": 612.29,
                    "word": " part",
                    "probability": 0.6396484375
                },
                {
                    "start": 612.29,
                    "end": 612.47,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 612.47,
                    "end": 612.87,
                    "word": " input",
                    "probability": 0.796875
                },
                {
                    "start": 612.87,
                    "end": 613.39,
                    "word": " A1",
                    "probability": 0.88134765625
                },
                {
                    "start": 613.39,
                    "end": 613.67,
                    "word": " is",
                    "probability": 0.88427734375
                },
                {
                    "start": 613.67,
                    "end": 614.09,
                    "word": " here",
                    "probability": 0.80712890625
                },
                {
                    "start": 614.09,
                    "end": 614.25,
                    "word": ",",
                    "probability": 0.8193359375
                },
                {
                    "start": 614.25,
                    "end": 614.53,
                    "word": " A2",
                    "probability": 0.98486328125
                },
                {
                    "start": 614.53,
                    "end": 614.79,
                    "word": " is",
                    "probability": 0.89697265625
                },
                {
                    "start": 614.79,
                    "end": 615.09,
                    "word": " here",
                    "probability": 0.81201171875
                },
                {
                    "start": 615.09,
                    "end": 615.15,
                    "word": ",",
                    "probability": 0.9501953125
                }
            ]
        },
        {
            "id": 97,
            "seek": 63280,
            "start": 615.15,
            "end": 623.69,
            "text": " A3 is here, and so on. So this way, if an adversary wants to attack the outputs, they need to",
            "tokens": [
                50740,
                316,
                18,
                307,
                510,
                11,
                293,
                370,
                322,
                13,
                407,
                341,
                636,
                11,
                498,
                364,
                48222,
                2738,
                281,
                2690,
                264,
                23930,
                11,
                436,
                643,
                281,
                51168
            ],
            "temperature": 0.0,
            "avg_logprob": -0.1833964583247599,
            "compression_ratio": 1.668103448275862,
            "no_speech_prob": 0.0025634765625,
            "words": [
                {
                    "start": 615.15,
                    "end": 615.57,
                    "word": " A3",
                    "probability": 0.990234375
                },
                {
                    "start": 615.57,
                    "end": 615.85,
                    "word": " is",
                    "probability": 0.89990234375
                },
                {
                    "start": 615.85,
                    "end": 616.07,
                    "word": " here",
                    "probability": 0.81298828125
                },
                {
                    "start": 616.07,
                    "end": 616.15,
                    "word": ",",
                    "probability": 0.5048828125
                },
                {
                    "start": 616.15,
                    "end": 616.23,
                    "word": " and",
                    "probability": 0.8916015625
                },
                {
                    "start": 616.23,
                    "end": 616.45,
                    "word": " so",
                    "probability": 0.8984375
                },
                {
                    "start": 616.45,
                    "end": 616.71,
                    "word": " on",
                    "probability": 0.90576171875
                },
                {
                    "start": 616.71,
                    "end": 617.03,
                    "word": ".",
                    "probability": 0.9755859375
                },
                {
                    "start": 617.03,
                    "end": 617.23,
                    "word": " So",
                    "probability": 0.8623046875
                },
                {
                    "start": 617.23,
                    "end": 617.51,
                    "word": " this",
                    "probability": 0.82080078125
                },
                {
                    "start": 617.51,
                    "end": 617.77,
                    "word": " way",
                    "probability": 0.8955078125
                },
                {
                    "start": 617.77,
                    "end": 617.91,
                    "word": ",",
                    "probability": 0.7607421875
                },
                {
                    "start": 617.91,
                    "end": 618.17,
                    "word": " if",
                    "probability": 0.90087890625
                },
                {
                    "start": 618.17,
                    "end": 618.51,
                    "word": " an",
                    "probability": 0.89990234375
                },
                {
                    "start": 618.51,
                    "end": 618.97,
                    "word": " adversary",
                    "probability": 0.83154296875
                },
                {
                    "start": 618.97,
                    "end": 619.49,
                    "word": " wants",
                    "probability": 0.72900390625
                },
                {
                    "start": 619.49,
                    "end": 619.71,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 619.71,
                    "end": 620.21,
                    "word": " attack",
                    "probability": 0.85400390625
                },
                {
                    "start": 620.21,
                    "end": 620.91,
                    "word": " the",
                    "probability": 0.80615234375
                },
                {
                    "start": 620.91,
                    "end": 621.63,
                    "word": " outputs",
                    "probability": 0.26513671875
                },
                {
                    "start": 621.63,
                    "end": 622.13,
                    "word": ",",
                    "probability": 0.89501953125
                },
                {
                    "start": 622.13,
                    "end": 622.31,
                    "word": " they",
                    "probability": 0.7265625
                },
                {
                    "start": 622.31,
                    "end": 623.45,
                    "word": " need",
                    "probability": 0.63671875
                },
                {
                    "start": 623.45,
                    "end": 623.69,
                    "word": " to",
                    "probability": 0.90869140625
                }
            ]
        },
        {
            "id": 98,
            "seek": 63280,
            "start": 623.69,
            "end": 629.85,
            "text": " be able to hack into all of the parties in order to make sense of the data. Otherwise, getting",
            "tokens": [
                51168,
                312,
                1075,
                281,
                10339,
                666,
                439,
                295,
                264,
                8265,
                294,
                1668,
                281,
                652,
                2020,
                295,
                264,
                1412,
                13,
                10328,
                11,
                1242,
                51480
            ],
            "temperature": 0.0,
            "avg_logprob": -0.1833964583247599,
            "compression_ratio": 1.668103448275862,
            "no_speech_prob": 0.0025634765625,
            "words": [
                {
                    "start": 623.69,
                    "end": 623.85,
                    "word": " be",
                    "probability": 0.8974609375
                },
                {
                    "start": 623.85,
                    "end": 624.13,
                    "word": " able",
                    "probability": 0.88916015625
                },
                {
                    "start": 624.13,
                    "end": 624.37,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 624.37,
                    "end": 624.69,
                    "word": " hack",
                    "probability": 0.89453125
                },
                {
                    "start": 624.69,
                    "end": 624.93,
                    "word": " into",
                    "probability": 0.7451171875
                },
                {
                    "start": 624.93,
                    "end": 625.33,
                    "word": " all",
                    "probability": 0.90087890625
                },
                {
                    "start": 625.33,
                    "end": 625.47,
                    "word": " of",
                    "probability": 0.890625
                },
                {
                    "start": 625.47,
                    "end": 625.61,
                    "word": " the",
                    "probability": 0.8037109375
                },
                {
                    "start": 625.61,
                    "end": 625.93,
                    "word": " parties",
                    "probability": 0.84765625
                },
                {
                    "start": 625.93,
                    "end": 626.27,
                    "word": " in",
                    "probability": 0.88037109375
                },
                {
                    "start": 626.27,
                    "end": 626.45,
                    "word": " order",
                    "probability": 0.82373046875
                },
                {
                    "start": 626.45,
                    "end": 626.85,
                    "word": " to",
                    "probability": 0.9091796875
                },
                {
                    "start": 626.85,
                    "end": 627.31,
                    "word": " make",
                    "probability": 0.87939453125
                },
                {
                    "start": 627.31,
                    "end": 627.63,
                    "word": " sense",
                    "probability": 0.73974609375
                },
                {
                    "start": 627.63,
                    "end": 627.85,
                    "word": " of",
                    "probability": 0.90673828125
                },
                {
                    "start": 627.85,
                    "end": 627.97,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 627.97,
                    "end": 628.25,
                    "word": " data",
                    "probability": 0.80419921875
                },
                {
                    "start": 628.25,
                    "end": 628.57,
                    "word": ".",
                    "probability": 0.8330078125
                },
                {
                    "start": 628.57,
                    "end": 628.97,
                    "word": " Otherwise",
                    "probability": 0.91015625
                },
                {
                    "start": 628.97,
                    "end": 629.41,
                    "word": ",",
                    "probability": 0.90576171875
                },
                {
                    "start": 629.41,
                    "end": 629.85,
                    "word": " getting",
                    "probability": 0.89501953125
                }
            ]
        },
        {
            "id": 99,
            "seek": 63280,
            "start": 629.85,
            "end": 637.11,
            "text": " information from one of the parties is meaningless. Another technique is federated machine learning.",
            "tokens": [
                51480,
                1589,
                490,
                472,
                295,
                264,
                8265,
                307,
                33232,
                13,
                3996,
                6532,
                307,
                38024,
                770,
                3479,
                2539,
                13,
                51836
            ],
            "temperature": 0.0,
            "avg_logprob": -0.1833964583247599,
            "compression_ratio": 1.668103448275862,
            "no_speech_prob": 0.0025634765625,
            "words": [
                {
                    "start": 629.85,
                    "end": 630.41,
                    "word": " information",
                    "probability": 0.75390625
                },
                {
                    "start": 630.41,
                    "end": 630.75,
                    "word": " from",
                    "probability": 0.82275390625
                },
                {
                    "start": 630.75,
                    "end": 630.95,
                    "word": " one",
                    "probability": 0.81884765625
                },
                {
                    "start": 630.95,
                    "end": 631.07,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 631.07,
                    "end": 631.23,
                    "word": " the",
                    "probability": 0.81396484375
                },
                {
                    "start": 631.23,
                    "end": 631.55,
                    "word": " parties",
                    "probability": 0.87451171875
                },
                {
                    "start": 631.55,
                    "end": 632.19,
                    "word": " is",
                    "probability": 0.90478515625
                },
                {
                    "start": 632.19,
                    "end": 632.67,
                    "word": " meaningless",
                    "probability": 0.82275390625
                },
                {
                    "start": 632.67,
                    "end": 633.31,
                    "word": ".",
                    "probability": 0.994140625
                },
                {
                    "start": 634.45,
                    "end": 634.77,
                    "word": " Another",
                    "probability": 0.84765625
                },
                {
                    "start": 634.77,
                    "end": 635.31,
                    "word": " technique",
                    "probability": 0.9033203125
                },
                {
                    "start": 635.31,
                    "end": 635.67,
                    "word": " is",
                    "probability": 0.8994140625
                },
                {
                    "start": 635.67,
                    "end": 636.17,
                    "word": " federated",
                    "probability": 0.809814453125
                },
                {
                    "start": 636.17,
                    "end": 636.47,
                    "word": " machine",
                    "probability": 0.69189453125
                },
                {
                    "start": 636.47,
                    "end": 636.89,
                    "word": " learning",
                    "probability": 0.853515625
                },
                {
                    "start": 636.89,
                    "end": 637.11,
                    "word": ".",
                    "probability": 0.990234375
                }
            ]
        },
        {
            "id": 100,
            "seek": 66280,
            "start": 637.11,
            "end": 643.47,
            "text": " Federated machine learning is very similar to the previous method. However, the difference is that",
            "tokens": [
                50364,
                45545,
                770,
                3479,
                2539,
                307,
                588,
                2531,
                281,
                264,
                3894,
                3170,
                13,
                2908,
                11,
                264,
                2649,
                307,
                300,
                50700
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22202931804421508,
            "compression_ratio": 1.6535087719298245,
            "no_speech_prob": 0.0002875328063964844,
            "words": [
                {
                    "start": 637.11,
                    "end": 637.55,
                    "word": " Federated",
                    "probability": 0.76904296875
                },
                {
                    "start": 637.55,
                    "end": 637.85,
                    "word": " machine",
                    "probability": 0.498046875
                },
                {
                    "start": 637.85,
                    "end": 638.23,
                    "word": " learning",
                    "probability": 0.83447265625
                },
                {
                    "start": 638.23,
                    "end": 638.71,
                    "word": " is",
                    "probability": 0.8974609375
                },
                {
                    "start": 638.71,
                    "end": 639.15,
                    "word": " very",
                    "probability": 0.75244140625
                },
                {
                    "start": 639.15,
                    "end": 639.71,
                    "word": " similar",
                    "probability": 0.9013671875
                },
                {
                    "start": 639.71,
                    "end": 640.25,
                    "word": " to",
                    "probability": 0.89990234375
                },
                {
                    "start": 640.25,
                    "end": 640.45,
                    "word": " the",
                    "probability": 0.81591796875
                },
                {
                    "start": 640.45,
                    "end": 640.77,
                    "word": " previous",
                    "probability": 0.76318359375
                },
                {
                    "start": 640.77,
                    "end": 641.15,
                    "word": " method",
                    "probability": 0.82470703125
                },
                {
                    "start": 641.15,
                    "end": 641.35,
                    "word": ".",
                    "probability": 0.6328125
                },
                {
                    "start": 641.35,
                    "end": 641.77,
                    "word": " However",
                    "probability": 0.837890625
                },
                {
                    "start": 641.77,
                    "end": 641.99,
                    "word": ",",
                    "probability": 0.9462890625
                },
                {
                    "start": 641.99,
                    "end": 642.23,
                    "word": " the",
                    "probability": 0.80908203125
                },
                {
                    "start": 642.23,
                    "end": 642.63,
                    "word": " difference",
                    "probability": 0.71533203125
                },
                {
                    "start": 642.63,
                    "end": 643.13,
                    "word": " is",
                    "probability": 0.90185546875
                },
                {
                    "start": 643.13,
                    "end": 643.47,
                    "word": " that",
                    "probability": 0.88037109375
                }
            ]
        },
        {
            "id": 101,
            "seek": 66280,
            "start": 643.47,
            "end": 652.03,
            "text": " we have multiple sets of users, and these users all contribute into building a classifier.",
            "tokens": [
                50700,
                321,
                362,
                3866,
                6352,
                295,
                5022,
                11,
                293,
                613,
                5022,
                439,
                10586,
                666,
                2390,
                257,
                1508,
                9902,
                13,
                51096
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22202931804421508,
            "compression_ratio": 1.6535087719298245,
            "no_speech_prob": 0.0002875328063964844,
            "words": [
                {
                    "start": 643.47,
                    "end": 644.15,
                    "word": " we",
                    "probability": 0.830078125
                },
                {
                    "start": 644.15,
                    "end": 644.79,
                    "word": " have",
                    "probability": 0.87060546875
                },
                {
                    "start": 644.79,
                    "end": 645.23,
                    "word": " multiple",
                    "probability": 0.7548828125
                },
                {
                    "start": 645.23,
                    "end": 645.63,
                    "word": " sets",
                    "probability": 0.78466796875
                },
                {
                    "start": 645.63,
                    "end": 646.03,
                    "word": " of",
                    "probability": 0.9091796875
                },
                {
                    "start": 646.03,
                    "end": 646.51,
                    "word": " users",
                    "probability": 0.84130859375
                },
                {
                    "start": 646.51,
                    "end": 646.91,
                    "word": ",",
                    "probability": 0.302978515625
                },
                {
                    "start": 646.91,
                    "end": 647.57,
                    "word": " and",
                    "probability": 0.8876953125
                },
                {
                    "start": 647.57,
                    "end": 648.53,
                    "word": " these",
                    "probability": 0.70263671875
                },
                {
                    "start": 648.53,
                    "end": 648.91,
                    "word": " users",
                    "probability": 0.82763671875
                },
                {
                    "start": 648.91,
                    "end": 649.39,
                    "word": " all",
                    "probability": 0.8974609375
                },
                {
                    "start": 649.39,
                    "end": 649.89,
                    "word": " contribute",
                    "probability": 0.89453125
                },
                {
                    "start": 649.89,
                    "end": 650.37,
                    "word": " into",
                    "probability": 0.669921875
                },
                {
                    "start": 650.37,
                    "end": 650.73,
                    "word": " building",
                    "probability": 0.80908203125
                },
                {
                    "start": 650.97,
                    "end": 651.63,
                    "word": " a classifier",
                    "probability": 0.95849609375
                },
                {
                    "start": 651.63,
                    "end": 652.03,
                    "word": ".",
                    "probability": 0.978515625
                }
            ]
        },
        {
            "id": 102,
            "seek": 66280,
            "start": 652.75,
            "end": 659.25,
            "text": " The way that we handle the creation of the outputs is that we assign some sort of weights",
            "tokens": [
                51156,
                440,
                636,
                300,
                321,
                4813,
                264,
                8016,
                295,
                264,
                23930,
                307,
                300,
                321,
                6269,
                512,
                1333,
                295,
                17443,
                51492
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22202931804421508,
            "compression_ratio": 1.6535087719298245,
            "no_speech_prob": 0.0002875328063964844,
            "words": [
                {
                    "start": 652.75,
                    "end": 653.19,
                    "word": " The",
                    "probability": 0.7529296875
                },
                {
                    "start": 653.19,
                    "end": 653.47,
                    "word": " way",
                    "probability": 0.85009765625
                },
                {
                    "start": 653.47,
                    "end": 653.69,
                    "word": " that",
                    "probability": 0.8779296875
                },
                {
                    "start": 653.69,
                    "end": 654.01,
                    "word": " we",
                    "probability": 0.90673828125
                },
                {
                    "start": 654.01,
                    "end": 654.57,
                    "word": " handle",
                    "probability": 0.73095703125
                },
                {
                    "start": 654.57,
                    "end": 654.93,
                    "word": " the",
                    "probability": 0.8037109375
                },
                {
                    "start": 654.93,
                    "end": 655.49,
                    "word": " creation",
                    "probability": 0.7802734375
                },
                {
                    "start": 655.49,
                    "end": 655.75,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 655.75,
                    "end": 655.89,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 655.89,
                    "end": 656.37,
                    "word": " outputs",
                    "probability": 0.326171875
                },
                {
                    "start": 656.37,
                    "end": 656.89,
                    "word": " is",
                    "probability": 0.84521484375
                },
                {
                    "start": 656.89,
                    "end": 657.19,
                    "word": " that",
                    "probability": 0.880859375
                },
                {
                    "start": 657.19,
                    "end": 657.53,
                    "word": " we",
                    "probability": 0.8798828125
                },
                {
                    "start": 657.53,
                    "end": 657.97,
                    "word": " assign",
                    "probability": 0.8564453125
                },
                {
                    "start": 657.97,
                    "end": 658.25,
                    "word": " some",
                    "probability": 0.80419921875
                },
                {
                    "start": 658.25,
                    "end": 658.55,
                    "word": " sort",
                    "probability": 0.8125
                },
                {
                    "start": 658.55,
                    "end": 658.79,
                    "word": " of",
                    "probability": 0.9111328125
                },
                {
                    "start": 658.79,
                    "end": 659.25,
                    "word": " weights",
                    "probability": 0.646484375
                }
            ]
        },
        {
            "id": 103,
            "seek": 66280,
            "start": 659.25,
            "end": 665.95,
            "text": " and locally update the average of classifiers by assigning weights to each of the trained models.",
            "tokens": [
                51492,
                293,
                16143,
                5623,
                264,
                4274,
                295,
                1508,
                23463,
                538,
                49602,
                17443,
                281,
                1184,
                295,
                264,
                8895,
                5245,
                13,
                51796
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22202931804421508,
            "compression_ratio": 1.6535087719298245,
            "no_speech_prob": 0.0002875328063964844,
            "words": [
                {
                    "start": 659.25,
                    "end": 660.03,
                    "word": " and",
                    "probability": 0.70947265625
                },
                {
                    "start": 660.03,
                    "end": 660.51,
                    "word": " locally",
                    "probability": 0.8427734375
                },
                {
                    "start": 660.51,
                    "end": 661.11,
                    "word": " update",
                    "probability": 0.81201171875
                },
                {
                    "start": 661.11,
                    "end": 661.41,
                    "word": " the",
                    "probability": 0.73291015625
                },
                {
                    "start": 661.41,
                    "end": 662.01,
                    "word": " average",
                    "probability": 0.69921875
                },
                {
                    "start": 662.01,
                    "end": 662.23,
                    "word": " of",
                    "probability": 0.8857421875
                },
                {
                    "start": 662.23,
                    "end": 662.93,
                    "word": " classifiers",
                    "probability": 0.91796875
                },
                {
                    "start": 662.93,
                    "end": 663.19,
                    "word": " by",
                    "probability": 0.80908203125
                },
                {
                    "start": 663.19,
                    "end": 663.61,
                    "word": " assigning",
                    "probability": 0.84375
                },
                {
                    "start": 663.61,
                    "end": 663.95,
                    "word": " weights",
                    "probability": 0.6279296875
                },
                {
                    "start": 663.95,
                    "end": 664.15,
                    "word": " to",
                    "probability": 0.90283203125
                },
                {
                    "start": 664.15,
                    "end": 664.45,
                    "word": " each",
                    "probability": 0.8720703125
                },
                {
                    "start": 664.45,
                    "end": 664.67,
                    "word": " of",
                    "probability": 0.9072265625
                },
                {
                    "start": 664.67,
                    "end": 664.81,
                    "word": " the",
                    "probability": 0.6982421875
                },
                {
                    "start": 664.81,
                    "end": 665.23,
                    "word": " trained",
                    "probability": 0.77685546875
                },
                {
                    "start": 665.23,
                    "end": 665.75,
                    "word": " models",
                    "probability": 0.77734375
                },
                {
                    "start": 665.75,
                    "end": 665.95,
                    "word": ".",
                    "probability": 0.99658203125
                }
            ]
        },
        {
            "id": 104,
            "seek": 69280,
            "start": 667.83,
            "end": 672.55,
            "text": " The most important technique to preserve privacy is differential privacy.",
            "tokens": [
                50408,
                440,
                881,
                1021,
                6532,
                281,
                15665,
                11427,
                307,
                15756,
                11427,
                13,
                50620
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19772376543209877,
            "compression_ratio": 1.8743961352657006,
            "no_speech_prob": 0.0002911090850830078,
            "words": [
                {
                    "start": 667.83,
                    "end": 668.25,
                    "word": " The",
                    "probability": 0.7001953125
                },
                {
                    "start": 668.25,
                    "end": 668.49,
                    "word": " most",
                    "probability": 0.81689453125
                },
                {
                    "start": 668.49,
                    "end": 668.97,
                    "word": " important",
                    "probability": 0.77734375
                },
                {
                    "start": 668.97,
                    "end": 669.43,
                    "word": " technique",
                    "probability": 0.89794921875
                },
                {
                    "start": 669.43,
                    "end": 669.67,
                    "word": " to",
                    "probability": 0.904296875
                },
                {
                    "start": 669.67,
                    "end": 670.11,
                    "word": " preserve",
                    "probability": 0.7626953125
                },
                {
                    "start": 670.11,
                    "end": 670.61,
                    "word": " privacy",
                    "probability": 0.84814453125
                },
                {
                    "start": 670.61,
                    "end": 671.07,
                    "word": " is",
                    "probability": 0.8935546875
                },
                {
                    "start": 671.07,
                    "end": 671.49,
                    "word": " differential",
                    "probability": 0.61865234375
                },
                {
                    "start": 671.49,
                    "end": 672.13,
                    "word": " privacy",
                    "probability": 0.8212890625
                },
                {
                    "start": 672.13,
                    "end": 672.55,
                    "word": ".",
                    "probability": 0.9375
                }
            ]
        },
        {
            "id": 105,
            "seek": 69280,
            "start": 672.93,
            "end": 679.05,
            "text": " Basically, differential privacy means that it's a noise injection technique so that",
            "tokens": [
                50652,
                8537,
                11,
                15756,
                11427,
                1355,
                300,
                309,
                311,
                257,
                5658,
                22873,
                6532,
                370,
                300,
                50976
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19772376543209877,
            "compression_ratio": 1.8743961352657006,
            "no_speech_prob": 0.0002911090850830078,
            "words": [
                {
                    "start": 672.93,
                    "end": 673.39,
                    "word": " Basically",
                    "probability": 0.82080078125
                },
                {
                    "start": 673.39,
                    "end": 673.55,
                    "word": ",",
                    "probability": 0.59765625
                },
                {
                    "start": 673.55,
                    "end": 673.91,
                    "word": " differential",
                    "probability": 0.6826171875
                },
                {
                    "start": 673.91,
                    "end": 674.53,
                    "word": " privacy",
                    "probability": 0.81494140625
                },
                {
                    "start": 674.53,
                    "end": 675.09,
                    "word": " means",
                    "probability": 0.8798828125
                },
                {
                    "start": 675.09,
                    "end": 675.45,
                    "word": " that",
                    "probability": 0.861328125
                },
                {
                    "start": 675.45,
                    "end": 675.99,
                    "word": " it's",
                    "probability": 0.903564453125
                },
                {
                    "start": 676.45,
                    "end": 677.01,
                    "word": " a noise",
                    "probability": 0.87646484375
                },
                {
                    "start": 677.01,
                    "end": 677.53,
                    "word": " injection",
                    "probability": 0.6552734375
                },
                {
                    "start": 677.53,
                    "end": 678.17,
                    "word": " technique",
                    "probability": 0.845703125
                },
                {
                    "start": 678.17,
                    "end": 678.65,
                    "word": " so",
                    "probability": 0.64697265625
                },
                {
                    "start": 678.65,
                    "end": 679.05,
                    "word": " that",
                    "probability": 0.87451171875
                }
            ]
        },
        {
            "id": 106,
            "seek": 69280,
            "start": 679.05,
            "end": 685.55,
            "text": " the output of your model shouldn't reveal any information about the input of the model.",
            "tokens": [
                50976,
                264,
                5598,
                295,
                428,
                2316,
                4659,
                380,
                10658,
                604,
                1589,
                466,
                264,
                4846,
                295,
                264,
                2316,
                13,
                51260
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19772376543209877,
            "compression_ratio": 1.8743961352657006,
            "no_speech_prob": 0.0002911090850830078,
            "words": [
                {
                    "start": 679.05,
                    "end": 679.61,
                    "word": " the",
                    "probability": 0.80029296875
                },
                {
                    "start": 679.61,
                    "end": 680.17,
                    "word": " output",
                    "probability": 0.8505859375
                },
                {
                    "start": 680.17,
                    "end": 680.51,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 680.51,
                    "end": 680.93,
                    "word": " your",
                    "probability": 0.80517578125
                },
                {
                    "start": 680.93,
                    "end": 681.39,
                    "word": " model",
                    "probability": 0.76953125
                },
                {
                    "start": 681.39,
                    "end": 681.97,
                    "word": " shouldn't",
                    "probability": 0.9248046875
                },
                {
                    "start": 681.97,
                    "end": 682.35,
                    "word": " reveal",
                    "probability": 0.90185546875
                },
                {
                    "start": 682.35,
                    "end": 682.63,
                    "word": " any",
                    "probability": 0.81005859375
                },
                {
                    "start": 682.63,
                    "end": 683.25,
                    "word": " information",
                    "probability": 0.72119140625
                },
                {
                    "start": 683.25,
                    "end": 683.87,
                    "word": " about",
                    "probability": 0.814453125
                },
                {
                    "start": 683.87,
                    "end": 684.09,
                    "word": " the",
                    "probability": 0.80712890625
                },
                {
                    "start": 684.09,
                    "end": 684.37,
                    "word": " input",
                    "probability": 0.85498046875
                },
                {
                    "start": 684.37,
                    "end": 684.61,
                    "word": " of",
                    "probability": 0.90869140625
                },
                {
                    "start": 684.61,
                    "end": 684.75,
                    "word": " the",
                    "probability": 0.8056640625
                },
                {
                    "start": 684.75,
                    "end": 685.09,
                    "word": " model",
                    "probability": 0.81591796875
                },
                {
                    "start": 685.09,
                    "end": 685.55,
                    "word": ".",
                    "probability": 0.69140625
                }
            ]
        },
        {
            "id": 107,
            "seek": 69280,
            "start": 685.55,
            "end": 691.69,
            "text": " Whether your model is trained using the data of this person or the data of this person",
            "tokens": [
                51288,
                8503,
                428,
                2316,
                307,
                8895,
                1228,
                264,
                1412,
                295,
                341,
                954,
                420,
                264,
                1412,
                295,
                341,
                954,
                51608
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19772376543209877,
            "compression_ratio": 1.8743961352657006,
            "no_speech_prob": 0.0002911090850830078,
            "words": [
                {
                    "start": 685.55,
                    "end": 685.89,
                    "word": " Whether",
                    "probability": 0.79443359375
                },
                {
                    "start": 685.89,
                    "end": 686.27,
                    "word": " your",
                    "probability": 0.79541015625
                },
                {
                    "start": 686.27,
                    "end": 686.71,
                    "word": " model",
                    "probability": 0.82080078125
                },
                {
                    "start": 686.71,
                    "end": 687.11,
                    "word": " is",
                    "probability": 0.89892578125
                },
                {
                    "start": 687.11,
                    "end": 687.53,
                    "word": " trained",
                    "probability": 0.865234375
                },
                {
                    "start": 687.53,
                    "end": 688.07,
                    "word": " using",
                    "probability": 0.8046875
                },
                {
                    "start": 688.07,
                    "end": 688.33,
                    "word": " the",
                    "probability": 0.81201171875
                },
                {
                    "start": 688.33,
                    "end": 688.67,
                    "word": " data",
                    "probability": 0.814453125
                },
                {
                    "start": 688.67,
                    "end": 689.19,
                    "word": " of",
                    "probability": 0.90185546875
                },
                {
                    "start": 689.19,
                    "end": 689.45,
                    "word": " this",
                    "probability": 0.880859375
                },
                {
                    "start": 689.45,
                    "end": 689.89,
                    "word": " person",
                    "probability": 0.79296875
                },
                {
                    "start": 689.89,
                    "end": 690.21,
                    "word": " or",
                    "probability": 0.8427734375
                },
                {
                    "start": 690.21,
                    "end": 690.39,
                    "word": " the",
                    "probability": 0.80859375
                },
                {
                    "start": 690.39,
                    "end": 690.67,
                    "word": " data",
                    "probability": 0.81640625
                },
                {
                    "start": 690.67,
                    "end": 690.99,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 690.99,
                    "end": 691.23,
                    "word": " this",
                    "probability": 0.87255859375
                },
                {
                    "start": 691.23,
                    "end": 691.69,
                    "word": " person",
                    "probability": 0.80859375
                }
            ]
        },
        {
            "id": 108,
            "seek": 69280,
            "start": 691.69,
            "end": 695.11,
            "text": " shouldn't be revealed in the outcome of your algorithm.",
            "tokens": [
                51608,
                4659,
                380,
                312,
                9599,
                294,
                264,
                9700,
                295,
                428,
                9284,
                13,
                51760
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19772376543209877,
            "compression_ratio": 1.8743961352657006,
            "no_speech_prob": 0.0002911090850830078,
            "words": [
                {
                    "start": 691.69,
                    "end": 692.43,
                    "word": " shouldn't",
                    "probability": 0.823486328125
                },
                {
                    "start": 692.43,
                    "end": 692.59,
                    "word": " be",
                    "probability": 0.8955078125
                },
                {
                    "start": 692.59,
                    "end": 692.97,
                    "word": " revealed",
                    "probability": 0.88525390625
                },
                {
                    "start": 692.97,
                    "end": 693.19,
                    "word": " in",
                    "probability": 0.90966796875
                },
                {
                    "start": 693.19,
                    "end": 693.39,
                    "word": " the",
                    "probability": 0.8154296875
                },
                {
                    "start": 693.39,
                    "end": 693.71,
                    "word": " outcome",
                    "probability": 0.82421875
                },
                {
                    "start": 693.71,
                    "end": 694.23,
                    "word": " of",
                    "probability": 0.90771484375
                },
                {
                    "start": 694.23,
                    "end": 694.43,
                    "word": " your",
                    "probability": 0.794921875
                },
                {
                    "start": 694.43,
                    "end": 694.95,
                    "word": " algorithm",
                    "probability": 0.83837890625
                },
                {
                    "start": 694.95,
                    "end": 695.11,
                    "word": ".",
                    "probability": 0.998046875
                }
            ]
        },
        {
            "id": 109,
            "seek": 71366,
            "start": 697.23,
            "end": 704.27,
            "text": " It can be non-interactive or interactive. For example, in case of non-interactive,",
            "tokens": [
                50388,
                467,
                393,
                312,
                2107,
                12,
                5106,
                12596,
                420,
                15141,
                13,
                1171,
                1365,
                11,
                294,
                1389,
                295,
                2107,
                12,
                5106,
                12596,
                11,
                50720
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19014246148221634,
            "compression_ratio": 1.8194444444444444,
            "no_speech_prob": 0.0004162788391113281,
            "words": [
                {
                    "start": 697.23,
                    "end": 697.87,
                    "word": " It",
                    "probability": 0.5673828125
                },
                {
                    "start": 697.87,
                    "end": 698.21,
                    "word": " can",
                    "probability": 0.89453125
                },
                {
                    "start": 698.21,
                    "end": 698.61,
                    "word": " be",
                    "probability": 0.8955078125
                },
                {
                    "start": 698.61,
                    "end": 699.15,
                    "word": " non",
                    "probability": 0.873046875
                },
                {
                    "start": 699.15,
                    "end": 699.73,
                    "word": "-interactive",
                    "probability": 0.86865234375
                },
                {
                    "start": 699.73,
                    "end": 700.23,
                    "word": " or",
                    "probability": 0.87158203125
                },
                {
                    "start": 700.23,
                    "end": 700.83,
                    "word": " interactive",
                    "probability": 0.71435546875
                },
                {
                    "start": 700.83,
                    "end": 701.47,
                    "word": ".",
                    "probability": 0.76220703125
                },
                {
                    "start": 701.69,
                    "end": 701.87,
                    "word": " For",
                    "probability": 0.8916015625
                },
                {
                    "start": 701.87,
                    "end": 702.27,
                    "word": " example",
                    "probability": 0.9375
                },
                {
                    "start": 702.27,
                    "end": 702.43,
                    "word": ",",
                    "probability": 0.9296875
                },
                {
                    "start": 702.43,
                    "end": 702.51,
                    "word": " in",
                    "probability": 0.88134765625
                },
                {
                    "start": 702.51,
                    "end": 702.73,
                    "word": " case",
                    "probability": 0.69384765625
                },
                {
                    "start": 702.73,
                    "end": 702.91,
                    "word": " of",
                    "probability": 0.89990234375
                },
                {
                    "start": 702.91,
                    "end": 703.13,
                    "word": " non",
                    "probability": 0.83056640625
                },
                {
                    "start": 703.13,
                    "end": 704.03,
                    "word": "-interactive",
                    "probability": 0.94189453125
                },
                {
                    "start": 704.03,
                    "end": 704.27,
                    "word": ",",
                    "probability": 0.7685546875
                }
            ]
        },
        {
            "id": 110,
            "seek": 71366,
            "start": 704.27,
            "end": 710.71,
            "text": " we pre-compute the stats, the amount of noise, the amount of relationship between the input",
            "tokens": [
                50720,
                321,
                659,
                12,
                21541,
                1169,
                264,
                18152,
                11,
                264,
                2372,
                295,
                5658,
                11,
                264,
                2372,
                295,
                2480,
                1296,
                264,
                4846,
                51040
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19014246148221634,
            "compression_ratio": 1.8194444444444444,
            "no_speech_prob": 0.0004162788391113281,
            "words": [
                {
                    "start": 704.27,
                    "end": 704.59,
                    "word": " we",
                    "probability": 0.8740234375
                },
                {
                    "start": 704.59,
                    "end": 704.99,
                    "word": " pre",
                    "probability": 0.8662109375
                },
                {
                    "start": 704.99,
                    "end": 705.63,
                    "word": "-compute",
                    "probability": 0.7757161458333334
                },
                {
                    "start": 705.63,
                    "end": 705.95,
                    "word": " the",
                    "probability": 0.70166015625
                },
                {
                    "start": 705.95,
                    "end": 706.63,
                    "word": " stats",
                    "probability": 0.841796875
                },
                {
                    "start": 706.63,
                    "end": 707.05,
                    "word": ",",
                    "probability": 0.9443359375
                },
                {
                    "start": 707.05,
                    "end": 707.19,
                    "word": " the",
                    "probability": 0.8125
                },
                {
                    "start": 707.19,
                    "end": 707.59,
                    "word": " amount",
                    "probability": 0.83984375
                },
                {
                    "start": 707.59,
                    "end": 707.89,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 707.89,
                    "end": 708.27,
                    "word": " noise",
                    "probability": 0.87744140625
                },
                {
                    "start": 708.27,
                    "end": 708.37,
                    "word": ",",
                    "probability": 0.98583984375
                },
                {
                    "start": 708.37,
                    "end": 708.49,
                    "word": " the",
                    "probability": 0.81494140625
                },
                {
                    "start": 708.49,
                    "end": 708.79,
                    "word": " amount",
                    "probability": 0.83056640625
                },
                {
                    "start": 708.79,
                    "end": 708.97,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 708.97,
                    "end": 709.51,
                    "word": " relationship",
                    "probability": 0.7509765625
                },
                {
                    "start": 709.51,
                    "end": 710.11,
                    "word": " between",
                    "probability": 0.775390625
                },
                {
                    "start": 710.11,
                    "end": 710.31,
                    "word": " the",
                    "probability": 0.8046875
                },
                {
                    "start": 710.31,
                    "end": 710.71,
                    "word": " input",
                    "probability": 0.8369140625
                }
            ]
        },
        {
            "id": 111,
            "seek": 71366,
            "start": 711.35,
            "end": 717.97,
            "text": " features, and then inject noise to that. In case of interactive, the amount of noise is",
            "tokens": [
                51076,
                4122,
                11,
                293,
                550,
                10711,
                5658,
                281,
                300,
                13,
                682,
                1389,
                295,
                15141,
                11,
                264,
                2372,
                295,
                5658,
                307,
                51396
            ],
            "temperature": 0.0,
            "avg_logprob": -0.19014246148221634,
            "compression_ratio": 1.8194444444444444,
            "no_speech_prob": 0.0004162788391113281,
            "words": [
                {
                    "start": 711.35,
                    "end": 711.89,
                    "word": " features",
                    "probability": 0.72802734375
                },
                {
                    "start": 711.89,
                    "end": 712.15,
                    "word": ",",
                    "probability": 0.58984375
                },
                {
                    "start": 712.15,
                    "end": 712.43,
                    "word": " and",
                    "probability": 0.896484375
                },
                {
                    "start": 712.43,
                    "end": 712.65,
                    "word": " then",
                    "probability": 0.73388671875
                },
                {
                    "start": 712.65,
                    "end": 713.13,
                    "word": " inject",
                    "probability": 0.76171875
                },
                {
                    "start": 713.13,
                    "end": 713.41,
                    "word": " noise",
                    "probability": 0.8896484375
                },
                {
                    "start": 713.41,
                    "end": 713.55,
                    "word": " to",
                    "probability": 0.7421875
                },
                {
                    "start": 713.55,
                    "end": 713.75,
                    "word": " that",
                    "probability": 0.8486328125
                },
                {
                    "start": 713.75,
                    "end": 713.87,
                    "word": ".",
                    "probability": 0.99365234375
                },
                {
                    "start": 713.87,
                    "end": 713.99,
                    "word": " In",
                    "probability": 0.90771484375
                },
                {
                    "start": 713.99,
                    "end": 714.23,
                    "word": " case",
                    "probability": 0.7705078125
                },
                {
                    "start": 714.23,
                    "end": 714.41,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 714.41,
                    "end": 714.91,
                    "word": " interactive",
                    "probability": 0.75
                },
                {
                    "start": 714.91,
                    "end": 715.67,
                    "word": ",",
                    "probability": 0.9775390625
                },
                {
                    "start": 715.67,
                    "end": 716.27,
                    "word": " the",
                    "probability": 0.80712890625
                },
                {
                    "start": 716.27,
                    "end": 716.55,
                    "word": " amount",
                    "probability": 0.84912109375
                },
                {
                    "start": 716.55,
                    "end": 716.77,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 716.77,
                    "end": 717.29,
                    "word": " noise",
                    "probability": 0.90087890625
                },
                {
                    "start": 717.29,
                    "end": 717.97,
                    "word": " is",
                    "probability": 0.89990234375
                }
            ]
        },
        {
            "id": 112,
            "seek": 73726,
            "start": 719.27,
            "end": 723.77,
            "text": " injected based on user's requests.",
            "tokens": [
                50412,
                294,
                2884,
                349,
                292,
                2361,
                322,
                4195,
                311,
                12475,
                13,
                50608
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3107638839218352,
            "compression_ratio": 1.4078947368421053,
            "no_speech_prob": 0.0023956298828125,
            "words": [
                {
                    "start": 719.27,
                    "end": 720.11,
                    "word": " injected",
                    "probability": 0.6718902587890625
                },
                {
                    "start": 720.11,
                    "end": 720.47,
                    "word": " based",
                    "probability": 0.7841796875
                },
                {
                    "start": 720.47,
                    "end": 721.09,
                    "word": " on",
                    "probability": 0.90576171875
                },
                {
                    "start": 721.09,
                    "end": 722.35,
                    "word": " user's",
                    "probability": 0.605224609375
                },
                {
                    "start": 722.35,
                    "end": 722.93,
                    "word": " requests",
                    "probability": 0.7802734375
                },
                {
                    "start": 722.93,
                    "end": 723.77,
                    "word": ".",
                    "probability": 0.96875
                }
            ]
        },
        {
            "id": 113,
            "seek": 73726,
            "start": 725.35,
            "end": 731.91,
            "text": " What does differential privacy mean? It means that if person i change their information",
            "tokens": [
                50744,
                708,
                775,
                15756,
                11427,
                914,
                30,
                467,
                1355,
                300,
                498,
                954,
                741,
                1319,
                641,
                1589,
                51064
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3107638839218352,
            "compression_ratio": 1.4078947368421053,
            "no_speech_prob": 0.0023956298828125,
            "words": [
                {
                    "start": 725.35,
                    "end": 725.77,
                    "word": " What",
                    "probability": 0.055938720703125
                },
                {
                    "start": 725.77,
                    "end": 726.07,
                    "word": " does",
                    "probability": 0.79345703125
                },
                {
                    "start": 726.07,
                    "end": 726.93,
                    "word": " differential",
                    "probability": 0.216064453125
                },
                {
                    "start": 726.93,
                    "end": 727.57,
                    "word": " privacy",
                    "probability": 0.83349609375
                },
                {
                    "start": 727.57,
                    "end": 727.95,
                    "word": " mean",
                    "probability": 0.88671875
                },
                {
                    "start": 727.95,
                    "end": 728.15,
                    "word": "?",
                    "probability": 0.98046875
                },
                {
                    "start": 728.15,
                    "end": 728.39,
                    "word": " It",
                    "probability": 0.83251953125
                },
                {
                    "start": 728.39,
                    "end": 728.73,
                    "word": " means",
                    "probability": 0.87841796875
                },
                {
                    "start": 728.73,
                    "end": 729.07,
                    "word": " that",
                    "probability": 0.85546875
                },
                {
                    "start": 729.07,
                    "end": 729.65,
                    "word": " if",
                    "probability": 0.755859375
                },
                {
                    "start": 729.65,
                    "end": 730.07,
                    "word": " person",
                    "probability": 0.58544921875
                },
                {
                    "start": 730.07,
                    "end": 730.39,
                    "word": " i",
                    "probability": 0.6650390625
                },
                {
                    "start": 730.39,
                    "end": 730.91,
                    "word": " change",
                    "probability": 0.50244140625
                },
                {
                    "start": 730.91,
                    "end": 731.19,
                    "word": " their",
                    "probability": 0.85302734375
                },
                {
                    "start": 731.19,
                    "end": 731.91,
                    "word": " information",
                    "probability": 0.7158203125
                }
            ]
        },
        {
            "id": 114,
            "seek": 73726,
            "start": 732.51,
            "end": 741.57,
            "text": " from x i to x prime i, then the probability of their output should not change by that much.",
            "tokens": [
                51092,
                490,
                2031,
                741,
                281,
                2031,
                5835,
                741,
                11,
                550,
                264,
                8482,
                295,
                641,
                5598,
                820,
                406,
                1319,
                538,
                300,
                709,
                13,
                51512
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3107638839218352,
            "compression_ratio": 1.4078947368421053,
            "no_speech_prob": 0.0023956298828125,
            "words": [
                {
                    "start": 732.51,
                    "end": 732.93,
                    "word": " from",
                    "probability": 0.796875
                },
                {
                    "start": 732.93,
                    "end": 733.39,
                    "word": " x",
                    "probability": 0.45458984375
                },
                {
                    "start": 733.39,
                    "end": 733.85,
                    "word": " i",
                    "probability": 0.64404296875
                },
                {
                    "start": 733.85,
                    "end": 734.25,
                    "word": " to",
                    "probability": 0.8779296875
                },
                {
                    "start": 734.25,
                    "end": 734.65,
                    "word": " x",
                    "probability": 0.96484375
                },
                {
                    "start": 734.65,
                    "end": 735.91,
                    "word": " prime",
                    "probability": 0.6279296875
                },
                {
                    "start": 735.91,
                    "end": 736.35,
                    "word": " i",
                    "probability": 0.8720703125
                },
                {
                    "start": 736.35,
                    "end": 736.67,
                    "word": ",",
                    "probability": 0.63427734375
                },
                {
                    "start": 736.67,
                    "end": 737.13,
                    "word": " then",
                    "probability": 0.70849609375
                },
                {
                    "start": 737.13,
                    "end": 737.45,
                    "word": " the",
                    "probability": 0.79443359375
                },
                {
                    "start": 737.45,
                    "end": 737.95,
                    "word": " probability",
                    "probability": 0.8994140625
                },
                {
                    "start": 737.95,
                    "end": 738.37,
                    "word": " of",
                    "probability": 0.90771484375
                },
                {
                    "start": 738.37,
                    "end": 738.59,
                    "word": " their",
                    "probability": 0.88720703125
                },
                {
                    "start": 738.59,
                    "end": 739.15,
                    "word": " output",
                    "probability": 0.8134765625
                },
                {
                    "start": 739.15,
                    "end": 739.55,
                    "word": " should",
                    "probability": 0.93310546875
                },
                {
                    "start": 739.55,
                    "end": 739.83,
                    "word": " not",
                    "probability": 0.89404296875
                },
                {
                    "start": 739.83,
                    "end": 740.21,
                    "word": " change",
                    "probability": 0.84716796875
                },
                {
                    "start": 740.21,
                    "end": 740.41,
                    "word": " by",
                    "probability": 0.89453125
                },
                {
                    "start": 740.41,
                    "end": 740.69,
                    "word": " that",
                    "probability": 0.88037109375
                },
                {
                    "start": 740.69,
                    "end": 741.01,
                    "word": " much",
                    "probability": 0.814453125
                },
                {
                    "start": 741.01,
                    "end": 741.57,
                    "word": ".",
                    "probability": 0.97509765625
                }
            ]
        },
        {
            "id": 115,
            "seek": 76646,
            "start": 741.57,
            "end": 750.01,
            "text": " What does it mean again? It means that as a person, if one of my features is changing,",
            "tokens": [
                50364,
                708,
                775,
                309,
                914,
                797,
                30,
                467,
                1355,
                300,
                382,
                257,
                954,
                11,
                498,
                472,
                295,
                452,
                4122,
                307,
                4473,
                11,
                50776
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2064606687995825,
            "compression_ratio": 1.688073394495413,
            "no_speech_prob": 0.005535125732421875,
            "words": [
                {
                    "start": 741.57,
                    "end": 741.75,
                    "word": " What",
                    "probability": 0.017608642578125
                },
                {
                    "start": 741.75,
                    "end": 742.09,
                    "word": " does",
                    "probability": 0.8466796875
                },
                {
                    "start": 742.09,
                    "end": 742.25,
                    "word": " it",
                    "probability": 0.853515625
                },
                {
                    "start": 742.25,
                    "end": 742.51,
                    "word": " mean",
                    "probability": 0.9169921875
                },
                {
                    "start": 742.51,
                    "end": 742.97,
                    "word": " again",
                    "probability": 0.5390625
                },
                {
                    "start": 742.97,
                    "end": 743.33,
                    "word": "?",
                    "probability": 0.90478515625
                },
                {
                    "start": 743.33,
                    "end": 743.45,
                    "word": " It",
                    "probability": 0.8388671875
                },
                {
                    "start": 743.45,
                    "end": 743.81,
                    "word": " means",
                    "probability": 0.87646484375
                },
                {
                    "start": 743.81,
                    "end": 744.19,
                    "word": " that",
                    "probability": 0.8515625
                },
                {
                    "start": 744.19,
                    "end": 744.77,
                    "word": " as",
                    "probability": 0.708984375
                },
                {
                    "start": 745.09,
                    "end": 745.71,
                    "word": " a person",
                    "probability": 0.81201171875
                },
                {
                    "start": 745.71,
                    "end": 746.49,
                    "word": ",",
                    "probability": 0.72021484375
                },
                {
                    "start": 746.49,
                    "end": 746.85,
                    "word": " if",
                    "probability": 0.81787109375
                },
                {
                    "start": 746.85,
                    "end": 748.05,
                    "word": " one",
                    "probability": 0.77685546875
                },
                {
                    "start": 748.05,
                    "end": 748.21,
                    "word": " of",
                    "probability": 0.908203125
                },
                {
                    "start": 748.21,
                    "end": 748.41,
                    "word": " my",
                    "probability": 0.90625
                },
                {
                    "start": 748.41,
                    "end": 748.83,
                    "word": " features",
                    "probability": 0.71875
                },
                {
                    "start": 748.83,
                    "end": 749.21,
                    "word": " is",
                    "probability": 0.90087890625
                },
                {
                    "start": 749.21,
                    "end": 749.81,
                    "word": " changing",
                    "probability": 0.8115234375
                },
                {
                    "start": 749.81,
                    "end": 750.01,
                    "word": ",",
                    "probability": 0.300537109375
                }
            ]
        },
        {
            "id": 116,
            "seek": 76646,
            "start": 750.01,
            "end": 756.99,
            "text": " and as a person who is contributing into training a model, that change shouldn't cause a huge change",
            "tokens": [
                50776,
                293,
                382,
                257,
                954,
                567,
                307,
                19270,
                666,
                3097,
                257,
                2316,
                11,
                300,
                1319,
                4659,
                380,
                3082,
                257,
                2603,
                1319,
                51136
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2064606687995825,
            "compression_ratio": 1.688073394495413,
            "no_speech_prob": 0.005535125732421875,
            "words": [
                {
                    "start": 750.01,
                    "end": 750.11,
                    "word": " and",
                    "probability": 0.7275390625
                },
                {
                    "start": 750.11,
                    "end": 750.33,
                    "word": " as",
                    "probability": 0.87841796875
                },
                {
                    "start": 750.43,
                    "end": 750.67,
                    "word": " a person",
                    "probability": 0.80859375
                },
                {
                    "start": 750.67,
                    "end": 750.83,
                    "word": " who",
                    "probability": 0.79638671875
                },
                {
                    "start": 750.83,
                    "end": 750.97,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 750.97,
                    "end": 751.45,
                    "word": " contributing",
                    "probability": 0.876953125
                },
                {
                    "start": 751.45,
                    "end": 751.95,
                    "word": " into",
                    "probability": 0.5859375
                },
                {
                    "start": 751.95,
                    "end": 752.51,
                    "word": " training",
                    "probability": 0.8076171875
                },
                {
                    "start": 752.75,
                    "end": 753.09,
                    "word": " a model",
                    "probability": 0.8076171875
                },
                {
                    "start": 753.09,
                    "end": 753.63,
                    "word": ",",
                    "probability": 0.912109375
                },
                {
                    "start": 753.63,
                    "end": 753.91,
                    "word": " that",
                    "probability": 0.87890625
                },
                {
                    "start": 753.91,
                    "end": 754.55,
                    "word": " change",
                    "probability": 0.82080078125
                },
                {
                    "start": 754.55,
                    "end": 755.35,
                    "word": " shouldn't",
                    "probability": 0.904052734375
                },
                {
                    "start": 755.35,
                    "end": 755.79,
                    "word": " cause",
                    "probability": 0.80908203125
                },
                {
                    "start": 756.05,
                    "end": 756.47,
                    "word": " a huge",
                    "probability": 0.82177734375
                },
                {
                    "start": 756.47,
                    "end": 756.99,
                    "word": " change",
                    "probability": 0.85107421875
                }
            ]
        },
        {
            "id": 117,
            "seek": 76646,
            "start": 756.99,
            "end": 762.51,
            "text": " in the output. Because if that causes that huge change, it will reveal some information",
            "tokens": [
                51136,
                294,
                264,
                5598,
                13,
                1436,
                498,
                300,
                7700,
                300,
                2603,
                1319,
                11,
                309,
                486,
                10658,
                512,
                1589,
                51416
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2064606687995825,
            "compression_ratio": 1.688073394495413,
            "no_speech_prob": 0.005535125732421875,
            "words": [
                {
                    "start": 756.99,
                    "end": 757.35,
                    "word": " in",
                    "probability": 0.90673828125
                },
                {
                    "start": 757.35,
                    "end": 757.51,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 757.51,
                    "end": 758.01,
                    "word": " output",
                    "probability": 0.8349609375
                },
                {
                    "start": 758.01,
                    "end": 758.43,
                    "word": ".",
                    "probability": 0.471923828125
                },
                {
                    "start": 758.43,
                    "end": 758.85,
                    "word": " Because",
                    "probability": 0.81201171875
                },
                {
                    "start": 758.85,
                    "end": 759.07,
                    "word": " if",
                    "probability": 0.86572265625
                },
                {
                    "start": 759.07,
                    "end": 759.29,
                    "word": " that",
                    "probability": 0.876953125
                },
                {
                    "start": 759.29,
                    "end": 759.69,
                    "word": " causes",
                    "probability": 0.80078125
                },
                {
                    "start": 759.69,
                    "end": 759.99,
                    "word": " that",
                    "probability": 0.85693359375
                },
                {
                    "start": 759.99,
                    "end": 760.37,
                    "word": " huge",
                    "probability": 0.8408203125
                },
                {
                    "start": 760.37,
                    "end": 760.81,
                    "word": " change",
                    "probability": 0.84033203125
                },
                {
                    "start": 760.81,
                    "end": 760.85,
                    "word": ",",
                    "probability": 0.95068359375
                },
                {
                    "start": 760.85,
                    "end": 761.03,
                    "word": " it",
                    "probability": 0.90283203125
                },
                {
                    "start": 761.03,
                    "end": 761.23,
                    "word": " will",
                    "probability": 0.81591796875
                },
                {
                    "start": 761.23,
                    "end": 761.61,
                    "word": " reveal",
                    "probability": 0.8525390625
                },
                {
                    "start": 761.61,
                    "end": 761.93,
                    "word": " some",
                    "probability": 0.80859375
                },
                {
                    "start": 761.93,
                    "end": 762.51,
                    "word": " information",
                    "probability": 0.720703125
                }
            ]
        },
        {
            "id": 118,
            "seek": 76646,
            "start": 762.51,
            "end": 770.77,
            "text": " about this instance or observance as an input. For example, if I change the color of my hair",
            "tokens": [
                51416,
                466,
                341,
                5197,
                420,
                9951,
                719,
                382,
                364,
                4846,
                13,
                1171,
                1365,
                11,
                498,
                286,
                1319,
                264,
                2017,
                295,
                452,
                2578,
                51824
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2064606687995825,
            "compression_ratio": 1.688073394495413,
            "no_speech_prob": 0.005535125732421875,
            "words": [
                {
                    "start": 762.51,
                    "end": 763.19,
                    "word": " about",
                    "probability": 0.82080078125
                },
                {
                    "start": 763.19,
                    "end": 763.51,
                    "word": " this",
                    "probability": 0.83740234375
                },
                {
                    "start": 763.51,
                    "end": 764.03,
                    "word": " instance",
                    "probability": 0.9140625
                },
                {
                    "start": 764.03,
                    "end": 764.29,
                    "word": " or",
                    "probability": 0.84521484375
                },
                {
                    "start": 764.29,
                    "end": 765.13,
                    "word": " observance",
                    "probability": 0.83251953125
                },
                {
                    "start": 765.13,
                    "end": 765.91,
                    "word": " as",
                    "probability": 0.86279296875
                },
                {
                    "start": 765.91,
                    "end": 766.09,
                    "word": " an",
                    "probability": 0.8935546875
                },
                {
                    "start": 766.09,
                    "end": 766.47,
                    "word": " input",
                    "probability": 0.87060546875
                },
                {
                    "start": 766.47,
                    "end": 766.89,
                    "word": ".",
                    "probability": 0.98681640625
                },
                {
                    "start": 766.89,
                    "end": 767.07,
                    "word": " For",
                    "probability": 0.90380859375
                },
                {
                    "start": 767.07,
                    "end": 767.51,
                    "word": " example",
                    "probability": 0.93115234375
                },
                {
                    "start": 767.51,
                    "end": 768.11,
                    "word": ",",
                    "probability": 0.9873046875
                },
                {
                    "start": 768.11,
                    "end": 768.53,
                    "word": " if",
                    "probability": 0.8935546875
                },
                {
                    "start": 768.53,
                    "end": 769.41,
                    "word": " I",
                    "probability": 0.9921875
                },
                {
                    "start": 769.41,
                    "end": 769.73,
                    "word": " change",
                    "probability": 0.7919921875
                },
                {
                    "start": 769.73,
                    "end": 769.89,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 769.89,
                    "end": 770.11,
                    "word": " color",
                    "probability": 0.7109375
                },
                {
                    "start": 770.11,
                    "end": 770.27,
                    "word": " of",
                    "probability": 0.89892578125
                },
                {
                    "start": 770.27,
                    "end": 770.45,
                    "word": " my",
                    "probability": 0.88818359375
                },
                {
                    "start": 770.45,
                    "end": 770.77,
                    "word": " hair",
                    "probability": 0.87548828125
                }
            ]
        },
        {
            "id": 119,
            "seek": 78842,
            "start": 770.77,
            "end": 777.81,
            "text": " from brown to green, and that has a huge effect on the output of the model, then just by observing",
            "tokens": [
                50364,
                490,
                6292,
                281,
                3092,
                11,
                293,
                300,
                575,
                257,
                2603,
                1802,
                322,
                264,
                5598,
                295,
                264,
                2316,
                11,
                550,
                445,
                538,
                22107,
                50716
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17099472418637343,
            "compression_ratio": 1.4871794871794872,
            "no_speech_prob": 0.00020754337310791016,
            "words": [
                {
                    "start": 770.77,
                    "end": 770.97,
                    "word": " from",
                    "probability": 0.2578125
                },
                {
                    "start": 770.97,
                    "end": 771.51,
                    "word": " brown",
                    "probability": 0.77783203125
                },
                {
                    "start": 771.51,
                    "end": 771.97,
                    "word": " to",
                    "probability": 0.88720703125
                },
                {
                    "start": 771.97,
                    "end": 772.47,
                    "word": " green",
                    "probability": 0.89111328125
                },
                {
                    "start": 772.47,
                    "end": 772.75,
                    "word": ",",
                    "probability": 0.31005859375
                },
                {
                    "start": 772.75,
                    "end": 772.81,
                    "word": " and",
                    "probability": 0.8603515625
                },
                {
                    "start": 772.81,
                    "end": 773.05,
                    "word": " that",
                    "probability": 0.87353515625
                },
                {
                    "start": 773.05,
                    "end": 773.31,
                    "word": " has",
                    "probability": 0.88671875
                },
                {
                    "start": 773.43,
                    "end": 773.67,
                    "word": " a huge",
                    "probability": 0.81396484375
                },
                {
                    "start": 773.67,
                    "end": 774.11,
                    "word": " effect",
                    "probability": 0.8095703125
                },
                {
                    "start": 774.11,
                    "end": 774.35,
                    "word": " on",
                    "probability": 0.904296875
                },
                {
                    "start": 774.35,
                    "end": 774.49,
                    "word": " the",
                    "probability": 0.81005859375
                },
                {
                    "start": 774.49,
                    "end": 774.79,
                    "word": " output",
                    "probability": 0.77490234375
                },
                {
                    "start": 774.79,
                    "end": 774.99,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 774.99,
                    "end": 775.13,
                    "word": " the",
                    "probability": 0.81494140625
                },
                {
                    "start": 775.13,
                    "end": 775.47,
                    "word": " model",
                    "probability": 0.8154296875
                },
                {
                    "start": 775.47,
                    "end": 775.99,
                    "word": ",",
                    "probability": 0.7265625
                },
                {
                    "start": 775.99,
                    "end": 776.35,
                    "word": " then",
                    "probability": 0.72607421875
                },
                {
                    "start": 776.35,
                    "end": 777.11,
                    "word": " just",
                    "probability": 0.79345703125
                },
                {
                    "start": 777.11,
                    "end": 777.29,
                    "word": " by",
                    "probability": 0.9091796875
                },
                {
                    "start": 777.29,
                    "end": 777.81,
                    "word": " observing",
                    "probability": 0.67333984375
                }
            ]
        },
        {
            "id": 120,
            "seek": 78842,
            "start": 777.81,
            "end": 784.61,
            "text": " that change, an adversary will know that I changed the color of my hair. So using differential",
            "tokens": [
                50716,
                300,
                1319,
                11,
                364,
                48222,
                486,
                458,
                300,
                286,
                3105,
                264,
                2017,
                295,
                452,
                2578,
                13,
                407,
                1228,
                15756,
                51060
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17099472418637343,
            "compression_ratio": 1.4871794871794872,
            "no_speech_prob": 0.00020754337310791016,
            "words": [
                {
                    "start": 777.81,
                    "end": 778.13,
                    "word": " that",
                    "probability": 0.8818359375
                },
                {
                    "start": 778.13,
                    "end": 778.89,
                    "word": " change",
                    "probability": 0.83154296875
                },
                {
                    "start": 778.89,
                    "end": 779.69,
                    "word": ",",
                    "probability": 0.728515625
                },
                {
                    "start": 779.69,
                    "end": 779.95,
                    "word": " an",
                    "probability": 0.82568359375
                },
                {
                    "start": 779.95,
                    "end": 780.35,
                    "word": " adversary",
                    "probability": 0.81298828125
                },
                {
                    "start": 780.35,
                    "end": 780.81,
                    "word": " will",
                    "probability": 0.81982421875
                },
                {
                    "start": 780.81,
                    "end": 781.19,
                    "word": " know",
                    "probability": 0.8173828125
                },
                {
                    "start": 781.19,
                    "end": 781.43,
                    "word": " that",
                    "probability": 0.8759765625
                },
                {
                    "start": 781.43,
                    "end": 781.61,
                    "word": " I",
                    "probability": 0.99365234375
                },
                {
                    "start": 781.61,
                    "end": 781.97,
                    "word": " changed",
                    "probability": 0.7099609375
                },
                {
                    "start": 781.97,
                    "end": 782.19,
                    "word": " the",
                    "probability": 0.8125
                },
                {
                    "start": 782.19,
                    "end": 782.39,
                    "word": " color",
                    "probability": 0.7841796875
                },
                {
                    "start": 782.39,
                    "end": 782.59,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 782.59,
                    "end": 782.77,
                    "word": " my",
                    "probability": 0.90087890625
                },
                {
                    "start": 782.77,
                    "end": 783.09,
                    "word": " hair",
                    "probability": 0.87109375
                },
                {
                    "start": 783.09,
                    "end": 783.55,
                    "word": ".",
                    "probability": 0.95068359375
                },
                {
                    "start": 783.55,
                    "end": 783.73,
                    "word": " So",
                    "probability": 0.76123046875
                },
                {
                    "start": 783.73,
                    "end": 784.21,
                    "word": " using",
                    "probability": 0.63818359375
                },
                {
                    "start": 784.21,
                    "end": 784.61,
                    "word": " differential",
                    "probability": 0.67138671875
                }
            ]
        },
        {
            "id": 121,
            "seek": 78842,
            "start": 784.61,
            "end": 792.73,
            "text": " privacy, we want to make that change as minimal as possible. And how do we do that? By computing",
            "tokens": [
                51060,
                11427,
                11,
                321,
                528,
                281,
                652,
                300,
                1319,
                382,
                13206,
                382,
                1944,
                13,
                400,
                577,
                360,
                321,
                360,
                300,
                30,
                3146,
                15866,
                51472
            ],
            "temperature": 0.0,
            "avg_logprob": -0.17099472418637343,
            "compression_ratio": 1.4871794871794872,
            "no_speech_prob": 0.00020754337310791016,
            "words": [
                {
                    "start": 784.61,
                    "end": 785.19,
                    "word": " privacy",
                    "probability": 0.81884765625
                },
                {
                    "start": 785.19,
                    "end": 785.51,
                    "word": ",",
                    "probability": 0.8916015625
                },
                {
                    "start": 785.51,
                    "end": 785.65,
                    "word": " we",
                    "probability": 0.89501953125
                },
                {
                    "start": 785.65,
                    "end": 786.09,
                    "word": " want",
                    "probability": 0.81103515625
                },
                {
                    "start": 786.09,
                    "end": 786.55,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 786.55,
                    "end": 787.29,
                    "word": " make",
                    "probability": 0.8759765625
                },
                {
                    "start": 787.29,
                    "end": 787.59,
                    "word": " that",
                    "probability": 0.8798828125
                },
                {
                    "start": 787.59,
                    "end": 788.03,
                    "word": " change",
                    "probability": 0.85009765625
                },
                {
                    "start": 788.03,
                    "end": 788.41,
                    "word": " as",
                    "probability": 0.9130859375
                },
                {
                    "start": 788.41,
                    "end": 788.75,
                    "word": " minimal",
                    "probability": 0.779296875
                },
                {
                    "start": 788.75,
                    "end": 789.29,
                    "word": " as",
                    "probability": 0.90380859375
                },
                {
                    "start": 789.29,
                    "end": 789.85,
                    "word": " possible",
                    "probability": 0.84033203125
                },
                {
                    "start": 789.85,
                    "end": 790.49,
                    "word": ".",
                    "probability": 0.919921875
                },
                {
                    "start": 790.49,
                    "end": 790.73,
                    "word": " And",
                    "probability": 0.83203125
                },
                {
                    "start": 790.73,
                    "end": 790.95,
                    "word": " how",
                    "probability": 0.89404296875
                },
                {
                    "start": 790.95,
                    "end": 791.11,
                    "word": " do",
                    "probability": 0.9091796875
                },
                {
                    "start": 791.11,
                    "end": 791.23,
                    "word": " we",
                    "probability": 0.90185546875
                },
                {
                    "start": 791.23,
                    "end": 791.35,
                    "word": " do",
                    "probability": 0.90673828125
                },
                {
                    "start": 791.35,
                    "end": 791.69,
                    "word": " that",
                    "probability": 0.87841796875
                },
                {
                    "start": 791.69,
                    "end": 791.87,
                    "word": "?",
                    "probability": 0.908203125
                },
                {
                    "start": 791.87,
                    "end": 792.17,
                    "word": " By",
                    "probability": 0.9140625
                },
                {
                    "start": 792.17,
                    "end": 792.73,
                    "word": " computing",
                    "probability": 0.7080078125
                }
            ]
        },
        {
            "id": 122,
            "seek": 81790,
            "start": 792.73,
            "end": 799.05,
            "text": " what is the effect of that change and making that as a noise input into our model.",
            "tokens": [
                50376,
                437,
                307,
                264,
                1802,
                295,
                300,
                1319,
                293,
                1455,
                300,
                382,
                257,
                5658,
                4846,
                666,
                527,
                2316,
                13,
                50644
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20607639501492184,
            "compression_ratio": 1.6849315068493151,
            "no_speech_prob": 0.00658416748046875,
            "words": [
                {
                    "start": 792.73,
                    "end": 793.17,
                    "word": " what",
                    "probability": 0.02374267578125
                },
                {
                    "start": 793.17,
                    "end": 793.41,
                    "word": " is",
                    "probability": 0.8447265625
                },
                {
                    "start": 793.41,
                    "end": 793.53,
                    "word": " the",
                    "probability": 0.80419921875
                },
                {
                    "start": 793.53,
                    "end": 793.83,
                    "word": " effect",
                    "probability": 0.8544921875
                },
                {
                    "start": 793.83,
                    "end": 794.01,
                    "word": " of",
                    "probability": 0.8994140625
                },
                {
                    "start": 794.01,
                    "end": 794.19,
                    "word": " that",
                    "probability": 0.8740234375
                },
                {
                    "start": 794.19,
                    "end": 794.59,
                    "word": " change",
                    "probability": 0.84619140625
                },
                {
                    "start": 794.59,
                    "end": 794.93,
                    "word": " and",
                    "probability": 0.73095703125
                },
                {
                    "start": 794.93,
                    "end": 795.47,
                    "word": " making",
                    "probability": 0.7919921875
                },
                {
                    "start": 795.47,
                    "end": 796.07,
                    "word": " that",
                    "probability": 0.88037109375
                },
                {
                    "start": 796.07,
                    "end": 796.33,
                    "word": " as",
                    "probability": 0.87841796875
                },
                {
                    "start": 796.49,
                    "end": 796.79,
                    "word": " a noise",
                    "probability": 0.91455078125
                },
                {
                    "start": 796.79,
                    "end": 797.37,
                    "word": " input",
                    "probability": 0.87255859375
                },
                {
                    "start": 797.37,
                    "end": 797.73,
                    "word": " into",
                    "probability": 0.712890625
                },
                {
                    "start": 797.73,
                    "end": 798.03,
                    "word": " our",
                    "probability": 0.830078125
                },
                {
                    "start": 798.03,
                    "end": 798.37,
                    "word": " model",
                    "probability": 0.8095703125
                },
                {
                    "start": 798.37,
                    "end": 799.05,
                    "word": ".",
                    "probability": 0.947265625
                }
            ]
        },
        {
            "id": 123,
            "seek": 81790,
            "start": 799.87,
            "end": 807.37,
            "text": " So looking again at the differential privacy, you see here that we have two datasets. We call these",
            "tokens": [
                50728,
                407,
                1237,
                797,
                412,
                264,
                15756,
                11427,
                11,
                291,
                536,
                510,
                300,
                321,
                362,
                732,
                42856,
                13,
                492,
                818,
                613,
                51112
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20607639501492184,
            "compression_ratio": 1.6849315068493151,
            "no_speech_prob": 0.00658416748046875,
            "words": [
                {
                    "start": 799.87,
                    "end": 800.33,
                    "word": " So",
                    "probability": 0.80322265625
                },
                {
                    "start": 800.33,
                    "end": 800.65,
                    "word": " looking",
                    "probability": 0.591796875
                },
                {
                    "start": 800.65,
                    "end": 801.01,
                    "word": " again",
                    "probability": 0.86083984375
                },
                {
                    "start": 801.01,
                    "end": 801.21,
                    "word": " at",
                    "probability": 0.86767578125
                },
                {
                    "start": 801.21,
                    "end": 801.31,
                    "word": " the",
                    "probability": 0.78466796875
                },
                {
                    "start": 801.31,
                    "end": 801.63,
                    "word": " differential",
                    "probability": 0.67919921875
                },
                {
                    "start": 801.63,
                    "end": 802.27,
                    "word": " privacy",
                    "probability": 0.80810546875
                },
                {
                    "start": 802.27,
                    "end": 802.81,
                    "word": ",",
                    "probability": 0.82666015625
                },
                {
                    "start": 802.81,
                    "end": 802.99,
                    "word": " you",
                    "probability": 0.88818359375
                },
                {
                    "start": 802.99,
                    "end": 803.29,
                    "word": " see",
                    "probability": 0.80810546875
                },
                {
                    "start": 803.29,
                    "end": 803.69,
                    "word": " here",
                    "probability": 0.82373046875
                },
                {
                    "start": 803.69,
                    "end": 804.01,
                    "word": " that",
                    "probability": 0.8798828125
                },
                {
                    "start": 804.01,
                    "end": 804.27,
                    "word": " we",
                    "probability": 0.90869140625
                },
                {
                    "start": 804.27,
                    "end": 804.85,
                    "word": " have",
                    "probability": 0.880859375
                },
                {
                    "start": 804.85,
                    "end": 805.39,
                    "word": " two",
                    "probability": 0.79443359375
                },
                {
                    "start": 805.39,
                    "end": 806.01,
                    "word": " datasets",
                    "probability": 0.447021484375
                },
                {
                    "start": 806.01,
                    "end": 806.35,
                    "word": ".",
                    "probability": 0.79052734375
                },
                {
                    "start": 806.35,
                    "end": 806.47,
                    "word": " We",
                    "probability": 0.89697265625
                },
                {
                    "start": 806.47,
                    "end": 806.87,
                    "word": " call",
                    "probability": 0.81494140625
                },
                {
                    "start": 806.87,
                    "end": 807.37,
                    "word": " these",
                    "probability": 0.70556640625
                }
            ]
        },
        {
            "id": 124,
            "seek": 81790,
            "start": 807.37,
            "end": 814.69,
            "text": " neighbor datasets, meaning that all of the features in both datasets are actually equal,",
            "tokens": [
                51112,
                5987,
                42856,
                11,
                3620,
                300,
                439,
                295,
                264,
                4122,
                294,
                1293,
                42856,
                366,
                767,
                2681,
                11,
                51460
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20607639501492184,
            "compression_ratio": 1.6849315068493151,
            "no_speech_prob": 0.00658416748046875,
            "words": [
                {
                    "start": 807.37,
                    "end": 808.37,
                    "word": " neighbor",
                    "probability": 0.62939453125
                },
                {
                    "start": 808.37,
                    "end": 809.27,
                    "word": " datasets",
                    "probability": 0.853515625
                },
                {
                    "start": 809.27,
                    "end": 810.05,
                    "word": ",",
                    "probability": 0.76953125
                },
                {
                    "start": 810.05,
                    "end": 810.35,
                    "word": " meaning",
                    "probability": 0.82958984375
                },
                {
                    "start": 810.35,
                    "end": 810.77,
                    "word": " that",
                    "probability": 0.87548828125
                },
                {
                    "start": 810.77,
                    "end": 811.09,
                    "word": " all",
                    "probability": 0.8916015625
                },
                {
                    "start": 811.09,
                    "end": 811.25,
                    "word": " of",
                    "probability": 0.84375
                },
                {
                    "start": 811.25,
                    "end": 811.39,
                    "word": " the",
                    "probability": 0.8134765625
                },
                {
                    "start": 811.39,
                    "end": 811.87,
                    "word": " features",
                    "probability": 0.73291015625
                },
                {
                    "start": 811.87,
                    "end": 812.21,
                    "word": " in",
                    "probability": 0.9052734375
                },
                {
                    "start": 812.21,
                    "end": 812.45,
                    "word": " both",
                    "probability": 0.7939453125
                },
                {
                    "start": 812.45,
                    "end": 812.95,
                    "word": " datasets",
                    "probability": 0.7626953125
                },
                {
                    "start": 812.95,
                    "end": 813.51,
                    "word": " are",
                    "probability": 0.89208984375
                },
                {
                    "start": 813.51,
                    "end": 813.95,
                    "word": " actually",
                    "probability": 0.76953125
                },
                {
                    "start": 813.95,
                    "end": 814.53,
                    "word": " equal",
                    "probability": 0.76708984375
                },
                {
                    "start": 814.53,
                    "end": 814.69,
                    "word": ",",
                    "probability": 0.327392578125
                }
            ]
        },
        {
            "id": 125,
            "seek": 81790,
            "start": 814.69,
            "end": 822.21,
            "text": " except that one feature. In here, it is x3 and x3 prime. We want to make sure that when we change",
            "tokens": [
                51460,
                3993,
                300,
                472,
                4111,
                13,
                682,
                510,
                11,
                309,
                307,
                2031,
                18,
                293,
                2031,
                18,
                5835,
                13,
                492,
                528,
                281,
                652,
                988,
                300,
                562,
                321,
                1319,
                51832
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20607639501492184,
            "compression_ratio": 1.6849315068493151,
            "no_speech_prob": 0.00658416748046875,
            "words": [
                {
                    "start": 814.69,
                    "end": 814.97,
                    "word": " except",
                    "probability": 0.8427734375
                },
                {
                    "start": 814.97,
                    "end": 815.33,
                    "word": " that",
                    "probability": 0.859375
                },
                {
                    "start": 815.33,
                    "end": 815.95,
                    "word": " one",
                    "probability": 0.8154296875
                },
                {
                    "start": 815.95,
                    "end": 816.37,
                    "word": " feature",
                    "probability": 0.87841796875
                },
                {
                    "start": 816.37,
                    "end": 816.57,
                    "word": ".",
                    "probability": 0.63134765625
                },
                {
                    "start": 816.57,
                    "end": 816.77,
                    "word": " In",
                    "probability": 0.87255859375
                },
                {
                    "start": 816.77,
                    "end": 817.09,
                    "word": " here",
                    "probability": 0.8076171875
                },
                {
                    "start": 817.09,
                    "end": 817.21,
                    "word": ",",
                    "probability": 0.50341796875
                },
                {
                    "start": 817.21,
                    "end": 817.31,
                    "word": " it",
                    "probability": 0.8994140625
                },
                {
                    "start": 817.31,
                    "end": 817.51,
                    "word": " is",
                    "probability": 0.908203125
                },
                {
                    "start": 817.51,
                    "end": 818.09,
                    "word": " x3",
                    "probability": 0.65087890625
                },
                {
                    "start": 818.09,
                    "end": 818.47,
                    "word": " and",
                    "probability": 0.876953125
                },
                {
                    "start": 818.47,
                    "end": 819.01,
                    "word": " x3",
                    "probability": 0.991943359375
                },
                {
                    "start": 819.01,
                    "end": 819.31,
                    "word": " prime",
                    "probability": 0.62109375
                },
                {
                    "start": 819.31,
                    "end": 819.65,
                    "word": ".",
                    "probability": 0.9892578125
                },
                {
                    "start": 819.65,
                    "end": 819.91,
                    "word": " We",
                    "probability": 0.904296875
                },
                {
                    "start": 819.91,
                    "end": 820.11,
                    "word": " want",
                    "probability": 0.81787109375
                },
                {
                    "start": 820.11,
                    "end": 820.23,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 820.23,
                    "end": 820.43,
                    "word": " make",
                    "probability": 0.88134765625
                },
                {
                    "start": 820.43,
                    "end": 820.77,
                    "word": " sure",
                    "probability": 0.810546875
                },
                {
                    "start": 820.77,
                    "end": 821.19,
                    "word": " that",
                    "probability": 0.88427734375
                },
                {
                    "start": 821.19,
                    "end": 821.47,
                    "word": " when",
                    "probability": 0.85595703125
                },
                {
                    "start": 821.47,
                    "end": 821.71,
                    "word": " we",
                    "probability": 0.89013671875
                },
                {
                    "start": 821.71,
                    "end": 822.21,
                    "word": " change",
                    "probability": 0.833984375
                }
            ]
        },
        {
            "id": 126,
            "seek": 84778,
            "start": 822.21,
            "end": 830.79,
            "text": " the input data from x3 to x3 prime, the output that is generated by the model is changing",
            "tokens": [
                50364,
                264,
                4846,
                1412,
                490,
                2031,
                18,
                281,
                2031,
                18,
                5835,
                11,
                264,
                5598,
                300,
                307,
                10833,
                538,
                264,
                2316,
                307,
                4473,
                50796
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20034722603029675,
            "compression_ratio": 1.5541125541125542,
            "no_speech_prob": 0.0014524459838867188,
            "words": [
                {
                    "start": 822.21,
                    "end": 822.35,
                    "word": " the",
                    "probability": 0.24658203125
                },
                {
                    "start": 822.35,
                    "end": 822.59,
                    "word": " input",
                    "probability": 0.60546875
                },
                {
                    "start": 822.59,
                    "end": 822.95,
                    "word": " data",
                    "probability": 0.8115234375
                },
                {
                    "start": 822.95,
                    "end": 823.35,
                    "word": " from",
                    "probability": 0.75732421875
                },
                {
                    "start": 823.35,
                    "end": 823.85,
                    "word": " x3",
                    "probability": 0.6552734375
                },
                {
                    "start": 823.85,
                    "end": 824.13,
                    "word": " to",
                    "probability": 0.8955078125
                },
                {
                    "start": 824.13,
                    "end": 824.83,
                    "word": " x3",
                    "probability": 0.984130859375
                },
                {
                    "start": 824.83,
                    "end": 825.29,
                    "word": " prime",
                    "probability": 0.431640625
                },
                {
                    "start": 825.29,
                    "end": 825.75,
                    "word": ",",
                    "probability": 0.73095703125
                },
                {
                    "start": 825.75,
                    "end": 826.03,
                    "word": " the",
                    "probability": 0.80029296875
                },
                {
                    "start": 826.03,
                    "end": 826.69,
                    "word": " output",
                    "probability": 0.8037109375
                },
                {
                    "start": 826.69,
                    "end": 827.45,
                    "word": " that",
                    "probability": 0.7978515625
                },
                {
                    "start": 827.45,
                    "end": 827.61,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 827.61,
                    "end": 828.05,
                    "word": " generated",
                    "probability": 0.7783203125
                },
                {
                    "start": 828.05,
                    "end": 828.39,
                    "word": " by",
                    "probability": 0.89208984375
                },
                {
                    "start": 828.39,
                    "end": 829.11,
                    "word": " the",
                    "probability": 0.81005859375
                },
                {
                    "start": 829.11,
                    "end": 829.47,
                    "word": " model",
                    "probability": 0.81640625
                },
                {
                    "start": 829.47,
                    "end": 830.03,
                    "word": " is",
                    "probability": 0.89013671875
                },
                {
                    "start": 830.03,
                    "end": 830.79,
                    "word": " changing",
                    "probability": 0.77294921875
                }
            ]
        },
        {
            "id": 127,
            "seek": 84778,
            "start": 830.79,
            "end": 836.29,
            "text": " minimally. So as you see, the probability of generating a specific output in the first case",
            "tokens": [
                50796,
                4464,
                379,
                13,
                407,
                382,
                291,
                536,
                11,
                264,
                8482,
                295,
                17746,
                257,
                2685,
                5598,
                294,
                264,
                700,
                1389,
                51080
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20034722603029675,
            "compression_ratio": 1.5541125541125542,
            "no_speech_prob": 0.0014524459838867188,
            "words": [
                {
                    "start": 830.79,
                    "end": 831.33,
                    "word": " minimally",
                    "probability": 0.874267578125
                },
                {
                    "start": 831.33,
                    "end": 831.47,
                    "word": ".",
                    "probability": 0.8544921875
                },
                {
                    "start": 831.47,
                    "end": 831.57,
                    "word": " So",
                    "probability": 0.8291015625
                },
                {
                    "start": 831.57,
                    "end": 831.83,
                    "word": " as",
                    "probability": 0.7099609375
                },
                {
                    "start": 831.83,
                    "end": 831.99,
                    "word": " you",
                    "probability": 0.896484375
                },
                {
                    "start": 831.99,
                    "end": 832.29,
                    "word": " see",
                    "probability": 0.82373046875
                },
                {
                    "start": 832.29,
                    "end": 832.41,
                    "word": ",",
                    "probability": 0.457275390625
                },
                {
                    "start": 832.41,
                    "end": 832.51,
                    "word": " the",
                    "probability": 0.6904296875
                },
                {
                    "start": 832.51,
                    "end": 833.17,
                    "word": " probability",
                    "probability": 0.90673828125
                },
                {
                    "start": 833.17,
                    "end": 833.61,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 833.61,
                    "end": 834.07,
                    "word": " generating",
                    "probability": 0.80419921875
                },
                {
                    "start": 834.35,
                    "end": 834.69,
                    "word": " a specific",
                    "probability": 0.78173828125
                },
                {
                    "start": 834.69,
                    "end": 835.25,
                    "word": " output",
                    "probability": 0.8173828125
                },
                {
                    "start": 835.25,
                    "end": 835.49,
                    "word": " in",
                    "probability": 0.8671875
                },
                {
                    "start": 835.49,
                    "end": 835.67,
                    "word": " the",
                    "probability": 0.81787109375
                },
                {
                    "start": 835.67,
                    "end": 835.93,
                    "word": " first",
                    "probability": 0.8212890625
                },
                {
                    "start": 835.93,
                    "end": 836.29,
                    "word": " case",
                    "probability": 0.7822265625
                }
            ]
        },
        {
            "id": 128,
            "seek": 84778,
            "start": 836.29,
            "end": 843.93,
            "text": " to the second case is minimal, or epsilon. The next method is called PATE, private",
            "tokens": [
                51080,
                281,
                264,
                1150,
                1389,
                307,
                13206,
                11,
                420,
                17889,
                13,
                440,
                958,
                3170,
                307,
                1219,
                31485,
                36,
                11,
                4551,
                51452
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20034722603029675,
            "compression_ratio": 1.5541125541125542,
            "no_speech_prob": 0.0014524459838867188,
            "words": [
                {
                    "start": 836.29,
                    "end": 836.77,
                    "word": " to",
                    "probability": 0.85009765625
                },
                {
                    "start": 836.77,
                    "end": 837.01,
                    "word": " the",
                    "probability": 0.80859375
                },
                {
                    "start": 837.01,
                    "end": 837.29,
                    "word": " second",
                    "probability": 0.77978515625
                },
                {
                    "start": 837.29,
                    "end": 837.81,
                    "word": " case",
                    "probability": 0.8173828125
                },
                {
                    "start": 837.81,
                    "end": 838.35,
                    "word": " is",
                    "probability": 0.884765625
                },
                {
                    "start": 838.35,
                    "end": 838.67,
                    "word": " minimal",
                    "probability": 0.73681640625
                },
                {
                    "start": 838.67,
                    "end": 838.89,
                    "word": ",",
                    "probability": 0.4375
                },
                {
                    "start": 838.89,
                    "end": 839.17,
                    "word": " or",
                    "probability": 0.88671875
                },
                {
                    "start": 839.17,
                    "end": 839.65,
                    "word": " epsilon",
                    "probability": 0.73095703125
                },
                {
                    "start": 839.65,
                    "end": 840.37,
                    "word": ".",
                    "probability": 0.9921875
                },
                {
                    "start": 841.31,
                    "end": 841.85,
                    "word": " The",
                    "probability": 0.80712890625
                },
                {
                    "start": 841.85,
                    "end": 842.09,
                    "word": " next",
                    "probability": 0.85986328125
                },
                {
                    "start": 842.09,
                    "end": 842.35,
                    "word": " method",
                    "probability": 0.83984375
                },
                {
                    "start": 842.35,
                    "end": 842.59,
                    "word": " is",
                    "probability": 0.90771484375
                },
                {
                    "start": 842.59,
                    "end": 842.95,
                    "word": " called",
                    "probability": 0.81982421875
                },
                {
                    "start": 842.95,
                    "end": 843.41,
                    "word": " PATE",
                    "probability": 0.6170654296875
                },
                {
                    "start": 843.41,
                    "end": 843.55,
                    "word": ",",
                    "probability": 0.8076171875
                },
                {
                    "start": 843.55,
                    "end": 843.93,
                    "word": " private",
                    "probability": 0.440673828125
                }
            ]
        },
        {
            "id": 129,
            "seek": 84778,
            "start": 843.93,
            "end": 852.09,
            "text": " aggregation of teacher ensembles. In this case, we have several classifiers that use some data",
            "tokens": [
                51452,
                16743,
                399,
                295,
                5027,
                12567,
                2504,
                904,
                13,
                682,
                341,
                1389,
                11,
                321,
                362,
                2940,
                1508,
                23463,
                300,
                764,
                512,
                1412,
                51860
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20034722603029675,
            "compression_ratio": 1.5541125541125542,
            "no_speech_prob": 0.0014524459838867188,
            "words": [
                {
                    "start": 843.93,
                    "end": 844.57,
                    "word": " aggregation",
                    "probability": 0.90087890625
                },
                {
                    "start": 844.57,
                    "end": 844.75,
                    "word": " of",
                    "probability": 0.908203125
                },
                {
                    "start": 844.75,
                    "end": 845.07,
                    "word": " teacher",
                    "probability": 0.84716796875
                },
                {
                    "start": 845.07,
                    "end": 845.79,
                    "word": " ensembles",
                    "probability": 0.8723958333333334
                },
                {
                    "start": 845.79,
                    "end": 845.93,
                    "word": ".",
                    "probability": 0.98779296875
                },
                {
                    "start": 845.93,
                    "end": 846.13,
                    "word": " In",
                    "probability": 0.90625
                },
                {
                    "start": 846.13,
                    "end": 846.43,
                    "word": " this",
                    "probability": 0.8779296875
                },
                {
                    "start": 846.43,
                    "end": 846.79,
                    "word": " case",
                    "probability": 0.810546875
                },
                {
                    "start": 846.79,
                    "end": 847.23,
                    "word": ",",
                    "probability": 0.9267578125
                },
                {
                    "start": 847.23,
                    "end": 847.59,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 847.59,
                    "end": 848.25,
                    "word": " have",
                    "probability": 0.8798828125
                },
                {
                    "start": 848.25,
                    "end": 849.25,
                    "word": " several",
                    "probability": 0.84326171875
                },
                {
                    "start": 849.25,
                    "end": 850.09,
                    "word": " classifiers",
                    "probability": 0.93115234375
                },
                {
                    "start": 850.09,
                    "end": 850.89,
                    "word": " that",
                    "probability": 0.87451171875
                },
                {
                    "start": 850.89,
                    "end": 851.37,
                    "word": " use",
                    "probability": 0.80908203125
                },
                {
                    "start": 851.37,
                    "end": 851.65,
                    "word": " some",
                    "probability": 0.81787109375
                },
                {
                    "start": 851.65,
                    "end": 852.09,
                    "word": " data",
                    "probability": 0.81396484375
                }
            ]
        },
        {
            "id": 130,
            "seek": 87642,
            "start": 852.09,
            "end": 859.47,
            "text": " use some features in order to train a model. Then we have two sets of voters. It's like an",
            "tokens": [
                50412,
                764,
                512,
                4122,
                294,
                1668,
                281,
                3847,
                257,
                2316,
                13,
                1396,
                321,
                362,
                732,
                6352,
                295,
                14073,
                13,
                467,
                311,
                411,
                364,
                50728
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20551470237619737,
            "compression_ratio": 1.5851528384279476,
            "no_speech_prob": 0.0006422996520996094,
            "words": [
                {
                    "start": 852.09,
                    "end": 853.01,
                    "word": " use",
                    "probability": 0.0284271240234375
                },
                {
                    "start": 853.01,
                    "end": 853.67,
                    "word": " some",
                    "probability": 0.77978515625
                },
                {
                    "start": 853.67,
                    "end": 854.09,
                    "word": " features",
                    "probability": 0.689453125
                },
                {
                    "start": 854.09,
                    "end": 854.37,
                    "word": " in",
                    "probability": 0.8173828125
                },
                {
                    "start": 854.37,
                    "end": 854.67,
                    "word": " order",
                    "probability": 0.83154296875
                },
                {
                    "start": 854.67,
                    "end": 854.95,
                    "word": " to",
                    "probability": 0.90185546875
                },
                {
                    "start": 854.95,
                    "end": 855.27,
                    "word": " train",
                    "probability": 0.7802734375
                },
                {
                    "start": 855.45,
                    "end": 855.75,
                    "word": " a model",
                    "probability": 0.8330078125
                },
                {
                    "start": 855.75,
                    "end": 856.33,
                    "word": ".",
                    "probability": 0.85302734375
                },
                {
                    "start": 856.33,
                    "end": 856.59,
                    "word": " Then",
                    "probability": 0.77880859375
                },
                {
                    "start": 856.59,
                    "end": 856.87,
                    "word": " we",
                    "probability": 0.8173828125
                },
                {
                    "start": 856.87,
                    "end": 857.23,
                    "word": " have",
                    "probability": 0.88134765625
                },
                {
                    "start": 857.23,
                    "end": 857.47,
                    "word": " two",
                    "probability": 0.77978515625
                },
                {
                    "start": 857.47,
                    "end": 857.81,
                    "word": " sets",
                    "probability": 0.82861328125
                },
                {
                    "start": 857.81,
                    "end": 858.05,
                    "word": " of",
                    "probability": 0.9052734375
                },
                {
                    "start": 858.05,
                    "end": 858.41,
                    "word": " voters",
                    "probability": 0.751953125
                },
                {
                    "start": 858.41,
                    "end": 858.75,
                    "word": ".",
                    "probability": 0.67431640625
                },
                {
                    "start": 858.75,
                    "end": 858.89,
                    "word": " It's",
                    "probability": 0.924560546875
                },
                {
                    "start": 858.89,
                    "end": 859.21,
                    "word": " like",
                    "probability": 0.86083984375
                },
                {
                    "start": 859.21,
                    "end": 859.47,
                    "word": " an",
                    "probability": 0.85595703125
                }
            ]
        },
        {
            "id": 131,
            "seek": 87642,
            "start": 859.47,
            "end": 865.27,
            "text": " inversion programming type of technique. These voters choose the majority voting generated by",
            "tokens": [
                50728,
                43576,
                9410,
                2010,
                295,
                6532,
                13,
                1981,
                14073,
                2826,
                264,
                6286,
                10419,
                10833,
                538,
                51024
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20551470237619737,
            "compression_ratio": 1.5851528384279476,
            "no_speech_prob": 0.0006422996520996094,
            "words": [
                {
                    "start": 859.47,
                    "end": 859.89,
                    "word": " inversion",
                    "probability": 0.292724609375
                },
                {
                    "start": 859.89,
                    "end": 860.61,
                    "word": " programming",
                    "probability": 0.62841796875
                },
                {
                    "start": 860.61,
                    "end": 860.97,
                    "word": " type",
                    "probability": 0.91162109375
                },
                {
                    "start": 860.97,
                    "end": 861.19,
                    "word": " of",
                    "probability": 0.89599609375
                },
                {
                    "start": 861.19,
                    "end": 861.57,
                    "word": " technique",
                    "probability": 0.84716796875
                },
                {
                    "start": 861.57,
                    "end": 862.03,
                    "word": ".",
                    "probability": 0.96728515625
                },
                {
                    "start": 862.03,
                    "end": 862.33,
                    "word": " These",
                    "probability": 0.89208984375
                },
                {
                    "start": 862.33,
                    "end": 862.75,
                    "word": " voters",
                    "probability": 0.912109375
                },
                {
                    "start": 862.75,
                    "end": 863.31,
                    "word": " choose",
                    "probability": 0.81591796875
                },
                {
                    "start": 863.31,
                    "end": 863.49,
                    "word": " the",
                    "probability": 0.80908203125
                },
                {
                    "start": 863.49,
                    "end": 863.87,
                    "word": " majority",
                    "probability": 0.83740234375
                },
                {
                    "start": 863.87,
                    "end": 864.41,
                    "word": " voting",
                    "probability": 0.83251953125
                },
                {
                    "start": 864.41,
                    "end": 864.97,
                    "word": " generated",
                    "probability": 0.77685546875
                },
                {
                    "start": 864.97,
                    "end": 865.27,
                    "word": " by",
                    "probability": 0.90869140625
                }
            ]
        },
        {
            "id": 132,
            "seek": 87642,
            "start": 865.27,
            "end": 873.13,
            "text": " all of these classifiers. And if we have equal amount of votes, for example, in this case,",
            "tokens": [
                51024,
                439,
                295,
                613,
                1508,
                23463,
                13,
                400,
                498,
                321,
                362,
                2681,
                2372,
                295,
                12068,
                11,
                337,
                1365,
                11,
                294,
                341,
                1389,
                11,
                51408
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20551470237619737,
            "compression_ratio": 1.5851528384279476,
            "no_speech_prob": 0.0006422996520996094,
            "words": [
                {
                    "start": 865.27,
                    "end": 865.53,
                    "word": " all",
                    "probability": 0.8896484375
                },
                {
                    "start": 865.53,
                    "end": 865.71,
                    "word": " of",
                    "probability": 0.89404296875
                },
                {
                    "start": 865.71,
                    "end": 865.93,
                    "word": " these",
                    "probability": 0.70166015625
                },
                {
                    "start": 865.93,
                    "end": 866.75,
                    "word": " classifiers",
                    "probability": 0.940673828125
                },
                {
                    "start": 866.75,
                    "end": 867.43,
                    "word": ".",
                    "probability": 0.71826171875
                },
                {
                    "start": 867.59,
                    "end": 867.93,
                    "word": " And",
                    "probability": 0.8291015625
                },
                {
                    "start": 867.93,
                    "end": 868.61,
                    "word": " if",
                    "probability": 0.8837890625
                },
                {
                    "start": 868.61,
                    "end": 869.57,
                    "word": " we",
                    "probability": 0.8857421875
                },
                {
                    "start": 869.57,
                    "end": 870.05,
                    "word": " have",
                    "probability": 0.8798828125
                },
                {
                    "start": 870.05,
                    "end": 870.51,
                    "word": " equal",
                    "probability": 0.76953125
                },
                {
                    "start": 870.51,
                    "end": 871.23,
                    "word": " amount",
                    "probability": 0.78173828125
                },
                {
                    "start": 871.23,
                    "end": 871.47,
                    "word": " of",
                    "probability": 0.90576171875
                },
                {
                    "start": 871.47,
                    "end": 871.83,
                    "word": " votes",
                    "probability": 0.7998046875
                },
                {
                    "start": 871.83,
                    "end": 871.89,
                    "word": ",",
                    "probability": 0.802734375
                },
                {
                    "start": 871.89,
                    "end": 872.01,
                    "word": " for",
                    "probability": 0.890625
                },
                {
                    "start": 872.01,
                    "end": 872.37,
                    "word": " example",
                    "probability": 0.93701171875
                },
                {
                    "start": 872.37,
                    "end": 872.49,
                    "word": ",",
                    "probability": 0.75439453125
                },
                {
                    "start": 872.49,
                    "end": 872.61,
                    "word": " in",
                    "probability": 0.90380859375
                },
                {
                    "start": 872.61,
                    "end": 872.83,
                    "word": " this",
                    "probability": 0.87646484375
                },
                {
                    "start": 872.83,
                    "end": 873.07,
                    "word": " case",
                    "probability": 0.80908203125
                },
                {
                    "start": 873.07,
                    "end": 873.13,
                    "word": ",",
                    "probability": 0.572265625
                }
            ]
        },
        {
            "id": 133,
            "seek": 87642,
            "start": 873.13,
            "end": 880.73,
            "text": " we have two cancers and two healthy results, then we will calculate a Gaussian noise to",
            "tokens": [
                51408,
                321,
                362,
                732,
                31063,
                293,
                732,
                4627,
                3542,
                11,
                550,
                321,
                486,
                8873,
                257,
                39148,
                5658,
                281,
                51780
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20551470237619737,
            "compression_ratio": 1.5851528384279476,
            "no_speech_prob": 0.0006422996520996094,
            "words": [
                {
                    "start": 873.13,
                    "end": 873.27,
                    "word": " we",
                    "probability": 0.89892578125
                },
                {
                    "start": 873.27,
                    "end": 873.47,
                    "word": " have",
                    "probability": 0.87451171875
                },
                {
                    "start": 873.47,
                    "end": 874.21,
                    "word": " two",
                    "probability": 0.5390625
                },
                {
                    "start": 874.21,
                    "end": 874.61,
                    "word": " cancers",
                    "probability": 0.73095703125
                },
                {
                    "start": 874.61,
                    "end": 874.97,
                    "word": " and",
                    "probability": 0.88134765625
                },
                {
                    "start": 874.97,
                    "end": 875.17,
                    "word": " two",
                    "probability": 0.8212890625
                },
                {
                    "start": 875.17,
                    "end": 875.49,
                    "word": " healthy",
                    "probability": 0.65087890625
                },
                {
                    "start": 875.49,
                    "end": 876.09,
                    "word": " results",
                    "probability": 0.71826171875
                },
                {
                    "start": 876.09,
                    "end": 876.65,
                    "word": ",",
                    "probability": 0.5390625
                },
                {
                    "start": 876.65,
                    "end": 877.13,
                    "word": " then",
                    "probability": 0.70751953125
                },
                {
                    "start": 877.13,
                    "end": 877.61,
                    "word": " we",
                    "probability": 0.8828125
                },
                {
                    "start": 877.61,
                    "end": 877.81,
                    "word": " will",
                    "probability": 0.7548828125
                },
                {
                    "start": 877.81,
                    "end": 878.45,
                    "word": " calculate",
                    "probability": 0.8564453125
                },
                {
                    "start": 878.79,
                    "end": 879.17,
                    "word": " a Gaussian",
                    "probability": 0.88671875
                },
                {
                    "start": 879.17,
                    "end": 880.03,
                    "word": " noise",
                    "probability": 0.958984375
                },
                {
                    "start": 880.03,
                    "end": 880.73,
                    "word": " to",
                    "probability": 0.89453125
                }
            ]
        },
        {
            "id": 134,
            "seek": 90402,
            "start": 880.73,
            "end": 887.27,
            "text": " inject to the voters data. So basically, we again go back into our training data,",
            "tokens": [
                50380,
                10711,
                281,
                264,
                14073,
                1412,
                13,
                407,
                1936,
                11,
                321,
                797,
                352,
                646,
                666,
                527,
                3097,
                1412,
                11,
                50684
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2387152761220932,
            "compression_ratio": 1.7452830188679245,
            "no_speech_prob": 0.00452423095703125,
            "words": [
                {
                    "start": 880.73,
                    "end": 881.49,
                    "word": " inject",
                    "probability": 0.01448822021484375
                },
                {
                    "start": 881.49,
                    "end": 881.79,
                    "word": " to",
                    "probability": 0.80859375
                },
                {
                    "start": 881.79,
                    "end": 882.01,
                    "word": " the",
                    "probability": 0.77001953125
                },
                {
                    "start": 882.01,
                    "end": 882.27,
                    "word": " voters",
                    "probability": 0.72509765625
                },
                {
                    "start": 882.27,
                    "end": 882.81,
                    "word": " data",
                    "probability": 0.61279296875
                },
                {
                    "start": 882.81,
                    "end": 883.41,
                    "word": ".",
                    "probability": 0.84521484375
                },
                {
                    "start": 883.47,
                    "end": 883.55,
                    "word": " So",
                    "probability": 0.8564453125
                },
                {
                    "start": 883.55,
                    "end": 884.07,
                    "word": " basically",
                    "probability": 0.7470703125
                },
                {
                    "start": 884.07,
                    "end": 884.79,
                    "word": ",",
                    "probability": 0.546875
                },
                {
                    "start": 884.79,
                    "end": 884.91,
                    "word": " we",
                    "probability": 0.89501953125
                },
                {
                    "start": 884.91,
                    "end": 885.31,
                    "word": " again",
                    "probability": 0.88623046875
                },
                {
                    "start": 885.31,
                    "end": 885.53,
                    "word": " go",
                    "probability": 0.861328125
                },
                {
                    "start": 885.53,
                    "end": 885.83,
                    "word": " back",
                    "probability": 0.81640625
                },
                {
                    "start": 885.83,
                    "end": 886.05,
                    "word": " into",
                    "probability": 0.716796875
                },
                {
                    "start": 886.05,
                    "end": 886.29,
                    "word": " our",
                    "probability": 0.80712890625
                },
                {
                    "start": 886.29,
                    "end": 886.65,
                    "word": " training",
                    "probability": 0.775390625
                },
                {
                    "start": 886.65,
                    "end": 887.09,
                    "word": " data",
                    "probability": 0.78662109375
                },
                {
                    "start": 887.09,
                    "end": 887.27,
                    "word": ",",
                    "probability": 0.1317138671875
                }
            ]
        },
        {
            "id": 135,
            "seek": 90402,
            "start": 887.27,
            "end": 895.53,
            "text": " and we take a look and see that how much noise does each classifier needs to be injected into.",
            "tokens": [
                50684,
                293,
                321,
                747,
                257,
                574,
                293,
                536,
                300,
                577,
                709,
                5658,
                775,
                1184,
                1508,
                9902,
                2203,
                281,
                312,
                36967,
                666,
                13,
                51112
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2387152761220932,
            "compression_ratio": 1.7452830188679245,
            "no_speech_prob": 0.00452423095703125,
            "words": [
                {
                    "start": 887.27,
                    "end": 887.37,
                    "word": " and",
                    "probability": 0.8828125
                },
                {
                    "start": 887.37,
                    "end": 887.53,
                    "word": " we",
                    "probability": 0.8984375
                },
                {
                    "start": 887.53,
                    "end": 887.77,
                    "word": " take",
                    "probability": 0.82275390625
                },
                {
                    "start": 887.93,
                    "end": 888.21,
                    "word": " a look",
                    "probability": 0.91162109375
                },
                {
                    "start": 888.21,
                    "end": 888.45,
                    "word": " and",
                    "probability": 0.88330078125
                },
                {
                    "start": 888.45,
                    "end": 888.77,
                    "word": " see",
                    "probability": 0.81396484375
                },
                {
                    "start": 888.77,
                    "end": 889.03,
                    "word": " that",
                    "probability": 0.767578125
                },
                {
                    "start": 889.03,
                    "end": 889.33,
                    "word": " how",
                    "probability": 0.88671875
                },
                {
                    "start": 889.33,
                    "end": 889.63,
                    "word": " much",
                    "probability": 0.818359375
                },
                {
                    "start": 889.63,
                    "end": 890.05,
                    "word": " noise",
                    "probability": 0.8935546875
                },
                {
                    "start": 890.05,
                    "end": 890.49,
                    "word": " does",
                    "probability": 0.89013671875
                },
                {
                    "start": 890.49,
                    "end": 891.05,
                    "word": " each",
                    "probability": 0.85400390625
                },
                {
                    "start": 891.05,
                    "end": 893.01,
                    "word": " classifier",
                    "probability": 0.948486328125
                },
                {
                    "start": 893.01,
                    "end": 894.01,
                    "word": " needs",
                    "probability": 0.6953125
                },
                {
                    "start": 894.01,
                    "end": 894.19,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 894.19,
                    "end": 894.37,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 894.37,
                    "end": 894.73,
                    "word": " injected",
                    "probability": 0.658203125
                },
                {
                    "start": 894.73,
                    "end": 895.27,
                    "word": " into",
                    "probability": 0.732421875
                },
                {
                    "start": 895.27,
                    "end": 895.53,
                    "word": ".",
                    "probability": 0.74609375
                }
            ]
        },
        {
            "id": 136,
            "seek": 90402,
            "start": 895.53,
            "end": 902.77,
            "text": " Then again, we look at we inject that noise and we get the results off of this chart based on the",
            "tokens": [
                51112,
                1396,
                797,
                11,
                321,
                574,
                412,
                321,
                10711,
                300,
                5658,
                293,
                321,
                483,
                264,
                3542,
                766,
                295,
                341,
                6927,
                2361,
                322,
                264,
                51460
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2387152761220932,
            "compression_ratio": 1.7452830188679245,
            "no_speech_prob": 0.00452423095703125,
            "words": [
                {
                    "start": 895.53,
                    "end": 895.99,
                    "word": " Then",
                    "probability": 0.81884765625
                },
                {
                    "start": 895.99,
                    "end": 896.41,
                    "word": " again",
                    "probability": 0.86572265625
                },
                {
                    "start": 896.41,
                    "end": 896.57,
                    "word": ",",
                    "probability": 0.78759765625
                },
                {
                    "start": 896.57,
                    "end": 896.67,
                    "word": " we",
                    "probability": 0.89697265625
                },
                {
                    "start": 896.67,
                    "end": 896.89,
                    "word": " look",
                    "probability": 0.61474609375
                },
                {
                    "start": 896.89,
                    "end": 897.21,
                    "word": " at",
                    "probability": 0.91162109375
                },
                {
                    "start": 897.21,
                    "end": 897.61,
                    "word": " we",
                    "probability": 0.222412109375
                },
                {
                    "start": 897.61,
                    "end": 898.01,
                    "word": " inject",
                    "probability": 0.67529296875
                },
                {
                    "start": 898.01,
                    "end": 898.19,
                    "word": " that",
                    "probability": 0.65380859375
                },
                {
                    "start": 898.19,
                    "end": 898.53,
                    "word": " noise",
                    "probability": 0.89990234375
                },
                {
                    "start": 898.53,
                    "end": 898.83,
                    "word": " and",
                    "probability": 0.66357421875
                },
                {
                    "start": 898.83,
                    "end": 899.01,
                    "word": " we",
                    "probability": 0.884765625
                },
                {
                    "start": 899.01,
                    "end": 899.17,
                    "word": " get",
                    "probability": 0.90185546875
                },
                {
                    "start": 899.17,
                    "end": 899.31,
                    "word": " the",
                    "probability": 0.60009765625
                },
                {
                    "start": 899.31,
                    "end": 899.71,
                    "word": " results",
                    "probability": 0.67138671875
                },
                {
                    "start": 899.71,
                    "end": 900.09,
                    "word": " off",
                    "probability": 0.4267578125
                },
                {
                    "start": 900.09,
                    "end": 900.45,
                    "word": " of",
                    "probability": 0.841796875
                },
                {
                    "start": 900.45,
                    "end": 900.83,
                    "word": " this",
                    "probability": 0.444580078125
                },
                {
                    "start": 900.83,
                    "end": 901.79,
                    "word": " chart",
                    "probability": 0.60302734375
                },
                {
                    "start": 901.79,
                    "end": 902.47,
                    "word": " based",
                    "probability": 0.568359375
                },
                {
                    "start": 902.47,
                    "end": 902.67,
                    "word": " on",
                    "probability": 0.91162109375
                },
                {
                    "start": 902.67,
                    "end": 902.77,
                    "word": " the",
                    "probability": 0.8115234375
                }
            ]
        },
        {
            "id": 137,
            "seek": 90402,
            "start": 902.77,
            "end": 908.33,
            "text": " results. So based on the amount of noise injected, the model might generate the value cancer or",
            "tokens": [
                51460,
                3542,
                13,
                407,
                2361,
                322,
                264,
                2372,
                295,
                5658,
                36967,
                11,
                264,
                2316,
                1062,
                8460,
                264,
                2158,
                5592,
                420,
                51740
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2387152761220932,
            "compression_ratio": 1.7452830188679245,
            "no_speech_prob": 0.00452423095703125,
            "words": [
                {
                    "start": 902.77,
                    "end": 903.13,
                    "word": " results",
                    "probability": 0.67724609375
                },
                {
                    "start": 903.13,
                    "end": 903.49,
                    "word": ".",
                    "probability": 0.9736328125
                },
                {
                    "start": 903.49,
                    "end": 903.73,
                    "word": " So",
                    "probability": 0.89501953125
                },
                {
                    "start": 903.73,
                    "end": 904.05,
                    "word": " based",
                    "probability": 0.72802734375
                },
                {
                    "start": 904.05,
                    "end": 904.19,
                    "word": " on",
                    "probability": 0.90966796875
                },
                {
                    "start": 904.19,
                    "end": 904.31,
                    "word": " the",
                    "probability": 0.81591796875
                },
                {
                    "start": 904.31,
                    "end": 904.59,
                    "word": " amount",
                    "probability": 0.86328125
                },
                {
                    "start": 904.59,
                    "end": 904.77,
                    "word": " of",
                    "probability": 0.90380859375
                },
                {
                    "start": 904.77,
                    "end": 904.99,
                    "word": " noise",
                    "probability": 0.92431640625
                },
                {
                    "start": 904.99,
                    "end": 905.45,
                    "word": " injected",
                    "probability": 0.6083984375
                },
                {
                    "start": 905.45,
                    "end": 905.75,
                    "word": ",",
                    "probability": 0.77392578125
                },
                {
                    "start": 905.75,
                    "end": 905.87,
                    "word": " the",
                    "probability": 0.80029296875
                },
                {
                    "start": 905.87,
                    "end": 906.15,
                    "word": " model",
                    "probability": 0.78759765625
                },
                {
                    "start": 906.15,
                    "end": 906.61,
                    "word": " might",
                    "probability": 0.81494140625
                },
                {
                    "start": 906.61,
                    "end": 907.13,
                    "word": " generate",
                    "probability": 0.81494140625
                },
                {
                    "start": 907.13,
                    "end": 907.29,
                    "word": " the",
                    "probability": 0.677734375
                },
                {
                    "start": 907.29,
                    "end": 907.57,
                    "word": " value",
                    "probability": 0.8955078125
                },
                {
                    "start": 907.57,
                    "end": 908.01,
                    "word": " cancer",
                    "probability": 0.7109375
                },
                {
                    "start": 908.01,
                    "end": 908.33,
                    "word": " or",
                    "probability": 0.83056640625
                }
            ]
        },
        {
            "id": 138,
            "seek": 92936,
            "start": 908.33,
            "end": 915.43,
            "text": " might generate the value healthy. The important thing is that whatever the output is, for example,",
            "tokens": [
                50364,
                1062,
                8460,
                264,
                2158,
                4627,
                13,
                440,
                1021,
                551,
                307,
                300,
                2035,
                264,
                5598,
                307,
                11,
                337,
                1365,
                11,
                50716
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20399306217829385,
            "compression_ratio": 1.7488584474885844,
            "no_speech_prob": 0.0095977783203125,
            "words": [
                {
                    "start": 908.33,
                    "end": 908.57,
                    "word": " might",
                    "probability": 0.2529296875
                },
                {
                    "start": 908.57,
                    "end": 908.95,
                    "word": " generate",
                    "probability": 0.6552734375
                },
                {
                    "start": 908.95,
                    "end": 909.11,
                    "word": " the",
                    "probability": 0.455322265625
                },
                {
                    "start": 909.11,
                    "end": 909.37,
                    "word": " value",
                    "probability": 0.8916015625
                },
                {
                    "start": 909.37,
                    "end": 909.77,
                    "word": " healthy",
                    "probability": 0.465087890625
                },
                {
                    "start": 909.77,
                    "end": 910.09,
                    "word": ".",
                    "probability": 0.9091796875
                },
                {
                    "start": 910.09,
                    "end": 910.43,
                    "word": " The",
                    "probability": 0.78515625
                },
                {
                    "start": 910.43,
                    "end": 910.83,
                    "word": " important",
                    "probability": 0.7890625
                },
                {
                    "start": 910.83,
                    "end": 911.31,
                    "word": " thing",
                    "probability": 0.80810546875
                },
                {
                    "start": 911.31,
                    "end": 911.67,
                    "word": " is",
                    "probability": 0.9033203125
                },
                {
                    "start": 911.67,
                    "end": 912.03,
                    "word": " that",
                    "probability": 0.86328125
                },
                {
                    "start": 912.03,
                    "end": 913.33,
                    "word": " whatever",
                    "probability": 0.53857421875
                },
                {
                    "start": 913.33,
                    "end": 913.85,
                    "word": " the",
                    "probability": 0.78564453125
                },
                {
                    "start": 913.85,
                    "end": 914.45,
                    "word": " output",
                    "probability": 0.7529296875
                },
                {
                    "start": 914.45,
                    "end": 914.85,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 914.85,
                    "end": 914.95,
                    "word": ",",
                    "probability": 0.80126953125
                },
                {
                    "start": 914.95,
                    "end": 915.07,
                    "word": " for",
                    "probability": 0.8798828125
                },
                {
                    "start": 915.07,
                    "end": 915.31,
                    "word": " example",
                    "probability": 0.9248046875
                },
                {
                    "start": 915.31,
                    "end": 915.43,
                    "word": ",",
                    "probability": 0.54150390625
                }
            ]
        },
        {
            "id": 139,
            "seek": 92936,
            "start": 915.43,
            "end": 921.61,
            "text": " in this case, it is cancer, the adversary cannot correlate it back to the data that is trained",
            "tokens": [
                50716,
                294,
                341,
                1389,
                11,
                309,
                307,
                5592,
                11,
                264,
                48222,
                2644,
                48742,
                309,
                646,
                281,
                264,
                1412,
                300,
                307,
                8895,
                51044
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20399306217829385,
            "compression_ratio": 1.7488584474885844,
            "no_speech_prob": 0.0095977783203125,
            "words": [
                {
                    "start": 915.43,
                    "end": 915.51,
                    "word": " in",
                    "probability": 0.896484375
                },
                {
                    "start": 915.51,
                    "end": 915.73,
                    "word": " this",
                    "probability": 0.88134765625
                },
                {
                    "start": 915.73,
                    "end": 916.01,
                    "word": " case",
                    "probability": 0.81494140625
                },
                {
                    "start": 916.01,
                    "end": 916.09,
                    "word": ",",
                    "probability": 0.441650390625
                },
                {
                    "start": 916.09,
                    "end": 916.19,
                    "word": " it",
                    "probability": 0.89697265625
                },
                {
                    "start": 916.19,
                    "end": 916.37,
                    "word": " is",
                    "probability": 0.9013671875
                },
                {
                    "start": 916.37,
                    "end": 916.89,
                    "word": " cancer",
                    "probability": 0.85791015625
                },
                {
                    "start": 916.89,
                    "end": 917.37,
                    "word": ",",
                    "probability": 0.59521484375
                },
                {
                    "start": 917.37,
                    "end": 917.49,
                    "word": " the",
                    "probability": 0.80126953125
                },
                {
                    "start": 917.49,
                    "end": 917.99,
                    "word": " adversary",
                    "probability": 0.84375
                },
                {
                    "start": 917.99,
                    "end": 918.55,
                    "word": " cannot",
                    "probability": 0.75146484375
                },
                {
                    "start": 918.55,
                    "end": 919.13,
                    "word": " correlate",
                    "probability": 0.849609375
                },
                {
                    "start": 919.13,
                    "end": 919.47,
                    "word": " it",
                    "probability": 0.88623046875
                },
                {
                    "start": 919.47,
                    "end": 919.89,
                    "word": " back",
                    "probability": 0.8212890625
                },
                {
                    "start": 919.89,
                    "end": 920.15,
                    "word": " to",
                    "probability": 0.90234375
                },
                {
                    "start": 920.15,
                    "end": 920.35,
                    "word": " the",
                    "probability": 0.80908203125
                },
                {
                    "start": 920.35,
                    "end": 920.67,
                    "word": " data",
                    "probability": 0.82421875
                },
                {
                    "start": 920.67,
                    "end": 920.99,
                    "word": " that",
                    "probability": 0.86865234375
                },
                {
                    "start": 920.99,
                    "end": 921.17,
                    "word": " is",
                    "probability": 0.88330078125
                },
                {
                    "start": 921.17,
                    "end": 921.61,
                    "word": " trained",
                    "probability": 0.826171875
                }
            ]
        },
        {
            "id": 140,
            "seek": 92936,
            "start": 921.61,
            "end": 928.61,
            "text": " in order to get the value of cancer here. Because even if we don't have we have the value healthy",
            "tokens": [
                51044,
                294,
                1668,
                281,
                483,
                264,
                2158,
                295,
                5592,
                510,
                13,
                1436,
                754,
                498,
                321,
                500,
                380,
                362,
                321,
                362,
                264,
                2158,
                4627,
                51380
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20399306217829385,
            "compression_ratio": 1.7488584474885844,
            "no_speech_prob": 0.0095977783203125,
            "words": [
                {
                    "start": 921.61,
                    "end": 922.13,
                    "word": " in",
                    "probability": 0.892578125
                },
                {
                    "start": 922.13,
                    "end": 922.41,
                    "word": " order",
                    "probability": 0.83154296875
                },
                {
                    "start": 922.41,
                    "end": 922.77,
                    "word": " to",
                    "probability": 0.90380859375
                },
                {
                    "start": 922.77,
                    "end": 923.11,
                    "word": " get",
                    "probability": 0.8984375
                },
                {
                    "start": 923.11,
                    "end": 923.43,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 923.43,
                    "end": 923.83,
                    "word": " value",
                    "probability": 0.91943359375
                },
                {
                    "start": 923.83,
                    "end": 924.07,
                    "word": " of",
                    "probability": 0.9033203125
                },
                {
                    "start": 924.07,
                    "end": 924.51,
                    "word": " cancer",
                    "probability": 0.84716796875
                },
                {
                    "start": 924.51,
                    "end": 924.91,
                    "word": " here",
                    "probability": 0.8134765625
                },
                {
                    "start": 924.91,
                    "end": 925.01,
                    "word": ".",
                    "probability": 0.39501953125
                },
                {
                    "start": 925.01,
                    "end": 925.53,
                    "word": " Because",
                    "probability": 0.88671875
                },
                {
                    "start": 925.53,
                    "end": 925.87,
                    "word": " even",
                    "probability": 0.763671875
                },
                {
                    "start": 925.87,
                    "end": 926.33,
                    "word": " if",
                    "probability": 0.9052734375
                },
                {
                    "start": 926.33,
                    "end": 926.85,
                    "word": " we",
                    "probability": 0.89697265625
                },
                {
                    "start": 926.85,
                    "end": 927.15,
                    "word": " don't",
                    "probability": 0.7685546875
                },
                {
                    "start": 927.15,
                    "end": 927.47,
                    "word": " have",
                    "probability": 0.86474609375
                },
                {
                    "start": 927.47,
                    "end": 927.63,
                    "word": " we",
                    "probability": 0.1561279296875
                },
                {
                    "start": 927.63,
                    "end": 927.83,
                    "word": " have",
                    "probability": 0.88134765625
                },
                {
                    "start": 927.83,
                    "end": 927.95,
                    "word": " the",
                    "probability": 0.80859375
                },
                {
                    "start": 927.95,
                    "end": 928.21,
                    "word": " value",
                    "probability": 0.890625
                },
                {
                    "start": 928.21,
                    "end": 928.61,
                    "word": " healthy",
                    "probability": 0.62060546875
                }
            ]
        },
        {
            "id": 141,
            "seek": 92936,
            "start": 928.61,
            "end": 933.67,
            "text": " here by the noise injection, these might also generate the value cancer. So we don't really",
            "tokens": [
                51380,
                510,
                538,
                264,
                5658,
                22873,
                11,
                613,
                1062,
                611,
                8460,
                264,
                2158,
                5592,
                13,
                407,
                321,
                500,
                380,
                534,
                51632
            ],
            "temperature": 0.0,
            "avg_logprob": -0.20399306217829385,
            "compression_ratio": 1.7488584474885844,
            "no_speech_prob": 0.0095977783203125,
            "words": [
                {
                    "start": 928.61,
                    "end": 928.91,
                    "word": " here",
                    "probability": 0.7783203125
                },
                {
                    "start": 928.91,
                    "end": 929.17,
                    "word": " by",
                    "probability": 0.53173828125
                },
                {
                    "start": 929.17,
                    "end": 929.37,
                    "word": " the",
                    "probability": 0.80224609375
                },
                {
                    "start": 929.37,
                    "end": 929.59,
                    "word": " noise",
                    "probability": 0.8798828125
                },
                {
                    "start": 929.59,
                    "end": 930.03,
                    "word": " injection",
                    "probability": 0.68115234375
                },
                {
                    "start": 930.03,
                    "end": 930.23,
                    "word": ",",
                    "probability": 0.70361328125
                },
                {
                    "start": 930.23,
                    "end": 930.41,
                    "word": " these",
                    "probability": 0.552734375
                },
                {
                    "start": 930.41,
                    "end": 930.65,
                    "word": " might",
                    "probability": 0.77587890625
                },
                {
                    "start": 930.65,
                    "end": 931.01,
                    "word": " also",
                    "probability": 0.822265625
                },
                {
                    "start": 931.01,
                    "end": 931.43,
                    "word": " generate",
                    "probability": 0.80029296875
                },
                {
                    "start": 931.43,
                    "end": 931.55,
                    "word": " the",
                    "probability": 0.58740234375
                },
                {
                    "start": 931.55,
                    "end": 931.79,
                    "word": " value",
                    "probability": 0.9052734375
                },
                {
                    "start": 931.79,
                    "end": 932.23,
                    "word": " cancer",
                    "probability": 0.69580078125
                },
                {
                    "start": 932.23,
                    "end": 932.53,
                    "word": ".",
                    "probability": 0.94921875
                },
                {
                    "start": 932.53,
                    "end": 932.95,
                    "word": " So",
                    "probability": 0.8916015625
                },
                {
                    "start": 932.95,
                    "end": 933.23,
                    "word": " we",
                    "probability": 0.81640625
                },
                {
                    "start": 933.23,
                    "end": 933.41,
                    "word": " don't",
                    "probability": 0.945068359375
                },
                {
                    "start": 933.41,
                    "end": 933.67,
                    "word": " really",
                    "probability": 0.775390625
                }
            ]
        },
        {
            "id": 142,
            "seek": 95486,
            "start": 933.67,
            "end": 940.07,
            "text": " know is James Smith the cause of generating the result cancer or if someone similar to",
            "tokens": [
                50368,
                458,
                307,
                5678,
                8538,
                264,
                3082,
                295,
                17746,
                264,
                1874,
                5592,
                420,
                498,
                1580,
                2531,
                281,
                50680
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25097657144069674,
            "compression_ratio": 1.7298578199052133,
            "no_speech_prob": 0.044036865234375,
            "words": [
                {
                    "start": 933.67,
                    "end": 933.99,
                    "word": " know",
                    "probability": 0.1954345703125
                },
                {
                    "start": 933.99,
                    "end": 934.29,
                    "word": " is",
                    "probability": 0.83984375
                },
                {
                    "start": 934.29,
                    "end": 934.63,
                    "word": " James",
                    "probability": 0.50439453125
                },
                {
                    "start": 934.63,
                    "end": 935.09,
                    "word": " Smith",
                    "probability": 0.409912109375
                },
                {
                    "start": 935.09,
                    "end": 935.73,
                    "word": " the",
                    "probability": 0.09423828125
                },
                {
                    "start": 935.73,
                    "end": 936.13,
                    "word": " cause",
                    "probability": 0.80224609375
                },
                {
                    "start": 936.13,
                    "end": 936.41,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 936.41,
                    "end": 936.87,
                    "word": " generating",
                    "probability": 0.78369140625
                },
                {
                    "start": 936.87,
                    "end": 937.07,
                    "word": " the",
                    "probability": 0.80810546875
                },
                {
                    "start": 937.07,
                    "end": 937.35,
                    "word": " result",
                    "probability": 0.775390625
                },
                {
                    "start": 937.35,
                    "end": 937.83,
                    "word": " cancer",
                    "probability": 0.69873046875
                },
                {
                    "start": 937.83,
                    "end": 938.37,
                    "word": " or",
                    "probability": 0.7412109375
                },
                {
                    "start": 938.37,
                    "end": 938.91,
                    "word": " if",
                    "probability": 0.89306640625
                },
                {
                    "start": 938.91,
                    "end": 939.35,
                    "word": " someone",
                    "probability": 0.83349609375
                },
                {
                    "start": 939.35,
                    "end": 939.81,
                    "word": " similar",
                    "probability": 0.876953125
                },
                {
                    "start": 939.81,
                    "end": 940.07,
                    "word": " to",
                    "probability": 0.8955078125
                }
            ]
        },
        {
            "id": 143,
            "seek": 95486,
            "start": 940.07,
            "end": 946.67,
            "text": " James Smith in this training data set is contributing into that result. Here is the",
            "tokens": [
                50680,
                5678,
                8538,
                294,
                341,
                3097,
                1412,
                992,
                307,
                19270,
                666,
                300,
                1874,
                13,
                1692,
                307,
                264,
                51008
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25097657144069674,
            "compression_ratio": 1.7298578199052133,
            "no_speech_prob": 0.044036865234375,
            "words": [
                {
                    "start": 940.07,
                    "end": 940.35,
                    "word": " James",
                    "probability": 0.85400390625
                },
                {
                    "start": 940.35,
                    "end": 940.65,
                    "word": " Smith",
                    "probability": 0.71435546875
                },
                {
                    "start": 940.65,
                    "end": 940.93,
                    "word": " in",
                    "probability": 0.888671875
                },
                {
                    "start": 940.93,
                    "end": 941.13,
                    "word": " this",
                    "probability": 0.86572265625
                },
                {
                    "start": 941.13,
                    "end": 941.45,
                    "word": " training",
                    "probability": 0.849609375
                },
                {
                    "start": 941.45,
                    "end": 941.73,
                    "word": " data",
                    "probability": 0.49072265625
                },
                {
                    "start": 941.73,
                    "end": 942.13,
                    "word": " set",
                    "probability": 0.85791015625
                },
                {
                    "start": 942.13,
                    "end": 942.73,
                    "word": " is",
                    "probability": 0.888671875
                },
                {
                    "start": 942.73,
                    "end": 943.25,
                    "word": " contributing",
                    "probability": 0.85595703125
                },
                {
                    "start": 943.25,
                    "end": 943.67,
                    "word": " into",
                    "probability": 0.638671875
                },
                {
                    "start": 943.67,
                    "end": 943.93,
                    "word": " that",
                    "probability": 0.88134765625
                },
                {
                    "start": 943.93,
                    "end": 944.41,
                    "word": " result",
                    "probability": 0.7861328125
                },
                {
                    "start": 944.41,
                    "end": 945.05,
                    "word": ".",
                    "probability": 0.89794921875
                },
                {
                    "start": 945.69,
                    "end": 946.33,
                    "word": " Here",
                    "probability": 0.83447265625
                },
                {
                    "start": 946.33,
                    "end": 946.51,
                    "word": " is",
                    "probability": 0.90234375
                },
                {
                    "start": 946.51,
                    "end": 946.67,
                    "word": " the",
                    "probability": 0.7861328125
                }
            ]
        },
        {
            "id": 144,
            "seek": 95486,
            "start": 946.67,
            "end": 952.29,
            "text": " comparison of the defense methods in terms of what they're emphasizing on, what are they trying to",
            "tokens": [
                51008,
                9660,
                295,
                264,
                7654,
                7150,
                294,
                2115,
                295,
                437,
                436,
                434,
                45550,
                322,
                11,
                437,
                366,
                436,
                1382,
                281,
                51292
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25097657144069674,
            "compression_ratio": 1.7298578199052133,
            "no_speech_prob": 0.044036865234375,
            "words": [
                {
                    "start": 946.67,
                    "end": 947.11,
                    "word": " comparison",
                    "probability": 0.69921875
                },
                {
                    "start": 947.11,
                    "end": 947.43,
                    "word": " of",
                    "probability": 0.89892578125
                },
                {
                    "start": 947.43,
                    "end": 947.55,
                    "word": " the",
                    "probability": 0.80419921875
                },
                {
                    "start": 947.55,
                    "end": 947.95,
                    "word": " defense",
                    "probability": 0.7734375
                },
                {
                    "start": 947.95,
                    "end": 948.31,
                    "word": " methods",
                    "probability": 0.70361328125
                },
                {
                    "start": 948.31,
                    "end": 948.63,
                    "word": " in",
                    "probability": 0.89404296875
                },
                {
                    "start": 948.63,
                    "end": 948.97,
                    "word": " terms",
                    "probability": 0.77001953125
                },
                {
                    "start": 948.97,
                    "end": 949.47,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 949.47,
                    "end": 949.75,
                    "word": " what",
                    "probability": 0.87841796875
                },
                {
                    "start": 949.75,
                    "end": 949.99,
                    "word": " they're",
                    "probability": 0.722900390625
                },
                {
                    "start": 949.99,
                    "end": 950.55,
                    "word": " emphasizing",
                    "probability": 0.8046875
                },
                {
                    "start": 950.55,
                    "end": 951.15,
                    "word": " on",
                    "probability": 0.81689453125
                },
                {
                    "start": 951.15,
                    "end": 951.27,
                    "word": ",",
                    "probability": 0.427734375
                },
                {
                    "start": 951.27,
                    "end": 951.41,
                    "word": " what",
                    "probability": 0.8740234375
                },
                {
                    "start": 951.41,
                    "end": 951.59,
                    "word": " are",
                    "probability": 0.8876953125
                },
                {
                    "start": 951.59,
                    "end": 951.75,
                    "word": " they",
                    "probability": 0.72412109375
                },
                {
                    "start": 951.75,
                    "end": 952.07,
                    "word": " trying",
                    "probability": 0.80810546875
                },
                {
                    "start": 952.07,
                    "end": 952.29,
                    "word": " to",
                    "probability": 0.9091796875
                }
            ]
        },
        {
            "id": 145,
            "seek": 95486,
            "start": 952.29,
            "end": 959.17,
            "text": " protect, whether it is the data owner or the model protection, and their use cases in different",
            "tokens": [
                51292,
                2371,
                11,
                1968,
                309,
                307,
                264,
                1412,
                7289,
                420,
                264,
                2316,
                6334,
                11,
                293,
                641,
                764,
                3331,
                294,
                819,
                51644
            ],
            "temperature": 0.0,
            "avg_logprob": -0.25097657144069674,
            "compression_ratio": 1.7298578199052133,
            "no_speech_prob": 0.044036865234375,
            "words": [
                {
                    "start": 952.29,
                    "end": 952.77,
                    "word": " protect",
                    "probability": 0.751953125
                },
                {
                    "start": 952.77,
                    "end": 952.81,
                    "word": ",",
                    "probability": 0.71142578125
                },
                {
                    "start": 952.81,
                    "end": 952.99,
                    "word": " whether",
                    "probability": 0.81591796875
                },
                {
                    "start": 952.99,
                    "end": 953.25,
                    "word": " it",
                    "probability": 0.8974609375
                },
                {
                    "start": 953.25,
                    "end": 953.43,
                    "word": " is",
                    "probability": 0.8974609375
                },
                {
                    "start": 953.43,
                    "end": 953.59,
                    "word": " the",
                    "probability": 0.80712890625
                },
                {
                    "start": 953.59,
                    "end": 953.89,
                    "word": " data",
                    "probability": 0.79736328125
                },
                {
                    "start": 953.89,
                    "end": 954.25,
                    "word": " owner",
                    "probability": 0.76220703125
                },
                {
                    "start": 954.25,
                    "end": 954.53,
                    "word": " or",
                    "probability": 0.880859375
                },
                {
                    "start": 954.53,
                    "end": 954.71,
                    "word": " the",
                    "probability": 0.80126953125
                },
                {
                    "start": 954.71,
                    "end": 955.17,
                    "word": " model",
                    "probability": 0.80859375
                },
                {
                    "start": 955.17,
                    "end": 956.15,
                    "word": " protection",
                    "probability": 0.587890625
                },
                {
                    "start": 956.15,
                    "end": 956.53,
                    "word": ",",
                    "probability": 0.40087890625
                },
                {
                    "start": 956.53,
                    "end": 956.95,
                    "word": " and",
                    "probability": 0.8974609375
                },
                {
                    "start": 956.95,
                    "end": 957.73,
                    "word": " their",
                    "probability": 0.88037109375
                },
                {
                    "start": 957.73,
                    "end": 958.01,
                    "word": " use",
                    "probability": 0.8095703125
                },
                {
                    "start": 958.01,
                    "end": 958.35,
                    "word": " cases",
                    "probability": 0.80615234375
                },
                {
                    "start": 958.35,
                    "end": 958.73,
                    "word": " in",
                    "probability": 0.9052734375
                },
                {
                    "start": 958.73,
                    "end": 959.17,
                    "word": " different",
                    "probability": 0.6884765625
                }
            ]
        },
        {
            "id": 146,
            "seek": 98056,
            "start": 959.17,
            "end": 965.39,
            "text": " applications. Most importantly, most of these methods are really effective when they're used",
            "tokens": [
                50364,
                5821,
                13,
                4534,
                8906,
                11,
                881,
                295,
                613,
                7150,
                366,
                534,
                4942,
                562,
                436,
                434,
                1143,
                50672
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2514022581088237,
            "compression_ratio": 1.663677130044843,
            "no_speech_prob": 0.00830078125,
            "words": [
                {
                    "start": 959.17,
                    "end": 959.85,
                    "word": " applications",
                    "probability": 0.456787109375
                },
                {
                    "start": 959.85,
                    "end": 960.55,
                    "word": ".",
                    "probability": 0.935546875
                },
                {
                    "start": 960.91,
                    "end": 961.61,
                    "word": " Most",
                    "probability": 0.791015625
                },
                {
                    "start": 961.61,
                    "end": 962.31,
                    "word": " importantly",
                    "probability": 0.64453125
                },
                {
                    "start": 962.31,
                    "end": 962.57,
                    "word": ",",
                    "probability": 0.78076171875
                },
                {
                    "start": 962.57,
                    "end": 962.79,
                    "word": " most",
                    "probability": 0.81103515625
                },
                {
                    "start": 962.79,
                    "end": 962.99,
                    "word": " of",
                    "probability": 0.90625
                },
                {
                    "start": 962.99,
                    "end": 963.19,
                    "word": " these",
                    "probability": 0.7041015625
                },
                {
                    "start": 963.19,
                    "end": 963.51,
                    "word": " methods",
                    "probability": 0.783203125
                },
                {
                    "start": 963.51,
                    "end": 963.85,
                    "word": " are",
                    "probability": 0.88720703125
                },
                {
                    "start": 963.85,
                    "end": 964.11,
                    "word": " really",
                    "probability": 0.73388671875
                },
                {
                    "start": 964.11,
                    "end": 964.61,
                    "word": " effective",
                    "probability": 0.73779296875
                },
                {
                    "start": 964.61,
                    "end": 964.89,
                    "word": " when",
                    "probability": 0.72607421875
                },
                {
                    "start": 964.89,
                    "end": 965.11,
                    "word": " they're",
                    "probability": 0.665771484375
                },
                {
                    "start": 965.11,
                    "end": 965.39,
                    "word": " used",
                    "probability": 0.80029296875
                }
            ]
        },
        {
            "id": 147,
            "seek": 98056,
            "start": 965.39,
            "end": 970.57,
            "text": " in combination with each other, one from each category. For example, homomorphic encryption",
            "tokens": [
                50672,
                294,
                6562,
                365,
                1184,
                661,
                11,
                472,
                490,
                1184,
                7719,
                13,
                1171,
                1365,
                11,
                3655,
                32702,
                299,
                29575,
                50936
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2514022581088237,
            "compression_ratio": 1.663677130044843,
            "no_speech_prob": 0.00830078125,
            "words": [
                {
                    "start": 965.39,
                    "end": 965.61,
                    "word": " in",
                    "probability": 0.89501953125
                },
                {
                    "start": 965.61,
                    "end": 966.05,
                    "word": " combination",
                    "probability": 0.88037109375
                },
                {
                    "start": 966.05,
                    "end": 966.39,
                    "word": " with",
                    "probability": 0.81396484375
                },
                {
                    "start": 966.39,
                    "end": 966.63,
                    "word": " each",
                    "probability": 0.85400390625
                },
                {
                    "start": 966.63,
                    "end": 967.01,
                    "word": " other",
                    "probability": 0.78173828125
                },
                {
                    "start": 967.01,
                    "end": 967.15,
                    "word": ",",
                    "probability": 0.5390625
                },
                {
                    "start": 967.15,
                    "end": 967.39,
                    "word": " one",
                    "probability": 0.80224609375
                },
                {
                    "start": 967.39,
                    "end": 967.65,
                    "word": " from",
                    "probability": 0.81884765625
                },
                {
                    "start": 967.65,
                    "end": 967.91,
                    "word": " each",
                    "probability": 0.86083984375
                },
                {
                    "start": 967.91,
                    "end": 968.41,
                    "word": " category",
                    "probability": 0.826171875
                },
                {
                    "start": 968.41,
                    "end": 968.61,
                    "word": ".",
                    "probability": 0.56982421875
                },
                {
                    "start": 968.61,
                    "end": 968.69,
                    "word": " For",
                    "probability": 0.90087890625
                },
                {
                    "start": 968.69,
                    "end": 969.19,
                    "word": " example",
                    "probability": 0.92919921875
                },
                {
                    "start": 969.19,
                    "end": 969.71,
                    "word": ",",
                    "probability": 0.9267578125
                },
                {
                    "start": 969.71,
                    "end": 970.19,
                    "word": " homomorphic",
                    "probability": 0.89794921875
                },
                {
                    "start": 970.19,
                    "end": 970.57,
                    "word": " encryption",
                    "probability": 0.7958984375
                }
            ]
        },
        {
            "id": 148,
            "seek": 98056,
            "start": 970.57,
            "end": 976.79,
            "text": " and differential privacy or SMP and differential privacy, PATE and federated lightning plus",
            "tokens": [
                50936,
                293,
                15756,
                11427,
                420,
                318,
                12224,
                293,
                15756,
                11427,
                11,
                17718,
                13639,
                293,
                38024,
                770,
                16589,
                1804,
                51260
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2514022581088237,
            "compression_ratio": 1.663677130044843,
            "no_speech_prob": 0.00830078125,
            "words": [
                {
                    "start": 970.57,
                    "end": 970.89,
                    "word": " and",
                    "probability": 0.86767578125
                },
                {
                    "start": 970.89,
                    "end": 971.25,
                    "word": " differential",
                    "probability": 0.705078125
                },
                {
                    "start": 971.25,
                    "end": 971.91,
                    "word": " privacy",
                    "probability": 0.83740234375
                },
                {
                    "start": 971.91,
                    "end": 972.49,
                    "word": " or",
                    "probability": 0.59619140625
                },
                {
                    "start": 972.49,
                    "end": 973.09,
                    "word": " SMP",
                    "probability": 0.809326171875
                },
                {
                    "start": 973.09,
                    "end": 973.43,
                    "word": " and",
                    "probability": 0.841796875
                },
                {
                    "start": 973.43,
                    "end": 973.75,
                    "word": " differential",
                    "probability": 0.7138671875
                },
                {
                    "start": 973.75,
                    "end": 974.45,
                    "word": " privacy",
                    "probability": 0.8369140625
                },
                {
                    "start": 974.45,
                    "end": 974.75,
                    "word": ",",
                    "probability": 0.57763671875
                },
                {
                    "start": 974.75,
                    "end": 974.89,
                    "word": " PATE",
                    "probability": 0.44439697265625
                },
                {
                    "start": 974.89,
                    "end": 975.27,
                    "word": " and",
                    "probability": 0.70361328125
                },
                {
                    "start": 975.27,
                    "end": 975.77,
                    "word": " federated",
                    "probability": 0.859130859375
                },
                {
                    "start": 975.77,
                    "end": 976.29,
                    "word": " lightning",
                    "probability": 0.256103515625
                },
                {
                    "start": 976.29,
                    "end": 976.79,
                    "word": " plus",
                    "probability": 0.341552734375
                }
            ]
        },
        {
            "id": 149,
            "seek": 98056,
            "start": 976.79,
            "end": 984.87,
            "text": " homomorphic encryption. Here are the practical methods for privacy preservation. They are very",
            "tokens": [
                51260,
                3655,
                32702,
                299,
                29575,
                13,
                1692,
                366,
                264,
                8496,
                7150,
                337,
                11427,
                27257,
                13,
                814,
                366,
                588,
                51648
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2514022581088237,
            "compression_ratio": 1.663677130044843,
            "no_speech_prob": 0.00830078125,
            "words": [
                {
                    "start": 976.79,
                    "end": 977.77,
                    "word": " homomorphic",
                    "probability": 0.81201171875
                },
                {
                    "start": 977.77,
                    "end": 978.99,
                    "word": " encryption",
                    "probability": 0.7958984375
                },
                {
                    "start": 978.99,
                    "end": 979.69,
                    "word": ".",
                    "probability": 0.9609375
                },
                {
                    "start": 980.65,
                    "end": 981.35,
                    "word": " Here",
                    "probability": 0.83203125
                },
                {
                    "start": 981.35,
                    "end": 981.61,
                    "word": " are",
                    "probability": 0.89111328125
                },
                {
                    "start": 981.61,
                    "end": 981.77,
                    "word": " the",
                    "probability": 0.80810546875
                },
                {
                    "start": 981.77,
                    "end": 982.15,
                    "word": " practical",
                    "probability": 0.85205078125
                },
                {
                    "start": 982.15,
                    "end": 982.73,
                    "word": " methods",
                    "probability": 0.7998046875
                },
                {
                    "start": 982.73,
                    "end": 983.01,
                    "word": " for",
                    "probability": 0.89794921875
                },
                {
                    "start": 983.01,
                    "end": 983.37,
                    "word": " privacy",
                    "probability": 0.8359375
                },
                {
                    "start": 983.37,
                    "end": 983.91,
                    "word": " preservation",
                    "probability": 0.697265625
                },
                {
                    "start": 983.91,
                    "end": 984.33,
                    "word": ".",
                    "probability": 0.69873046875
                },
                {
                    "start": 984.33,
                    "end": 984.39,
                    "word": " They",
                    "probability": 0.73583984375
                },
                {
                    "start": 984.39,
                    "end": 984.57,
                    "word": " are",
                    "probability": 0.88916015625
                },
                {
                    "start": 984.57,
                    "end": 984.87,
                    "word": " very",
                    "probability": 0.744140625
                }
            ]
        },
        {
            "id": 150,
            "seek": 100830,
            "start": 984.87,
            "end": 992.53,
            "text": " easy to use. There are already packages that are generated for that like TensorFlow or PyTorch",
            "tokens": [
                50364,
                1858,
                281,
                764,
                13,
                821,
                366,
                1217,
                17401,
                300,
                366,
                10833,
                337,
                300,
                411,
                37624,
                420,
                9953,
                51,
                284,
                339,
                50744
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22323802541545096,
            "compression_ratio": 1.6977777777777778,
            "no_speech_prob": 0.010986328125,
            "words": [
                {
                    "start": 984.87,
                    "end": 984.97,
                    "word": " easy",
                    "probability": 0.1763916015625
                },
                {
                    "start": 984.97,
                    "end": 985.39,
                    "word": " to",
                    "probability": 0.8876953125
                },
                {
                    "start": 985.39,
                    "end": 985.89,
                    "word": " use",
                    "probability": 0.83447265625
                },
                {
                    "start": 985.89,
                    "end": 986.37,
                    "word": ".",
                    "probability": 0.6923828125
                },
                {
                    "start": 986.57,
                    "end": 986.71,
                    "word": " There",
                    "probability": 0.72998046875
                },
                {
                    "start": 986.71,
                    "end": 986.89,
                    "word": " are",
                    "probability": 0.8916015625
                },
                {
                    "start": 986.89,
                    "end": 987.31,
                    "word": " already",
                    "probability": 0.80859375
                },
                {
                    "start": 987.31,
                    "end": 988.51,
                    "word": " packages",
                    "probability": 0.82177734375
                },
                {
                    "start": 988.51,
                    "end": 988.83,
                    "word": " that",
                    "probability": 0.84521484375
                },
                {
                    "start": 988.83,
                    "end": 988.99,
                    "word": " are",
                    "probability": 0.8876953125
                },
                {
                    "start": 988.99,
                    "end": 989.43,
                    "word": " generated",
                    "probability": 0.76953125
                },
                {
                    "start": 989.43,
                    "end": 989.79,
                    "word": " for",
                    "probability": 0.875
                },
                {
                    "start": 989.79,
                    "end": 990.09,
                    "word": " that",
                    "probability": 0.87646484375
                },
                {
                    "start": 990.09,
                    "end": 990.39,
                    "word": " like",
                    "probability": 0.371337890625
                },
                {
                    "start": 990.39,
                    "end": 990.85,
                    "word": " TensorFlow",
                    "probability": 0.556640625
                },
                {
                    "start": 990.85,
                    "end": 991.73,
                    "word": " or",
                    "probability": 0.69482421875
                },
                {
                    "start": 991.73,
                    "end": 992.53,
                    "word": " PyTorch",
                    "probability": 0.9281005859375
                }
            ]
        },
        {
            "id": 151,
            "seek": 100830,
            "start": 992.53,
            "end": 999.67,
            "text": " or IBM privacy package. We are going to show an example of how to use one of these packages in a",
            "tokens": [
                50744,
                420,
                23487,
                11427,
                7372,
                13,
                492,
                366,
                516,
                281,
                855,
                364,
                1365,
                295,
                577,
                281,
                764,
                472,
                295,
                613,
                17401,
                294,
                257,
                51080
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22323802541545096,
            "compression_ratio": 1.6977777777777778,
            "no_speech_prob": 0.010986328125,
            "words": [
                {
                    "start": 992.53,
                    "end": 992.83,
                    "word": " or",
                    "probability": 0.84912109375
                },
                {
                    "start": 992.83,
                    "end": 993.23,
                    "word": " IBM",
                    "probability": 0.90673828125
                },
                {
                    "start": 993.23,
                    "end": 993.73,
                    "word": " privacy",
                    "probability": 0.41552734375
                },
                {
                    "start": 993.73,
                    "end": 994.27,
                    "word": " package",
                    "probability": 0.9052734375
                },
                {
                    "start": 994.27,
                    "end": 994.53,
                    "word": ".",
                    "probability": 0.95556640625
                },
                {
                    "start": 994.53,
                    "end": 994.87,
                    "word": " We",
                    "probability": 0.91064453125
                },
                {
                    "start": 994.87,
                    "end": 995.05,
                    "word": " are",
                    "probability": 0.88671875
                },
                {
                    "start": 995.05,
                    "end": 995.31,
                    "word": " going",
                    "probability": 0.8525390625
                },
                {
                    "start": 995.31,
                    "end": 995.49,
                    "word": " to",
                    "probability": 0.90869140625
                },
                {
                    "start": 995.49,
                    "end": 995.69,
                    "word": " show",
                    "probability": 0.87841796875
                },
                {
                    "start": 995.69,
                    "end": 995.87,
                    "word": " an",
                    "probability": 0.89453125
                },
                {
                    "start": 995.87,
                    "end": 996.39,
                    "word": " example",
                    "probability": 0.92724609375
                },
                {
                    "start": 996.39,
                    "end": 996.91,
                    "word": " of",
                    "probability": 0.90087890625
                },
                {
                    "start": 996.91,
                    "end": 997.33,
                    "word": " how",
                    "probability": 0.90283203125
                },
                {
                    "start": 997.33,
                    "end": 997.59,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 997.59,
                    "end": 997.79,
                    "word": " use",
                    "probability": 0.83251953125
                },
                {
                    "start": 997.79,
                    "end": 997.99,
                    "word": " one",
                    "probability": 0.8095703125
                },
                {
                    "start": 997.99,
                    "end": 998.15,
                    "word": " of",
                    "probability": 0.90576171875
                },
                {
                    "start": 998.15,
                    "end": 998.35,
                    "word": " these",
                    "probability": 0.69677734375
                },
                {
                    "start": 998.35,
                    "end": 998.79,
                    "word": " packages",
                    "probability": 0.85400390625
                },
                {
                    "start": 998.79,
                    "end": 999.21,
                    "word": " in",
                    "probability": 0.89306640625
                },
                {
                    "start": 999.33,
                    "end": 999.67,
                    "word": " a real",
                    "probability": 0.89306640625
                }
            ]
        },
        {
            "id": 152,
            "seek": 100830,
            "start": 999.67,
            "end": 1008.27,
            "text": " real world sample. In this demo, we're going to show how you're going to implement a solution",
            "tokens": [
                51080,
                957,
                1002,
                6889,
                13,
                682,
                341,
                10723,
                11,
                321,
                434,
                516,
                281,
                855,
                577,
                291,
                434,
                516,
                281,
                4445,
                257,
                3827,
                51520
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22323802541545096,
            "compression_ratio": 1.6977777777777778,
            "no_speech_prob": 0.010986328125,
            "words": [
                {
                    "start": 999.67,
                    "end": 1000.41,
                    "word": " world",
                    "probability": 0.41796875
                },
                {
                    "start": 1000.41,
                    "end": 1000.97,
                    "word": " sample",
                    "probability": 0.62548828125
                },
                {
                    "start": 1000.97,
                    "end": 1001.45,
                    "word": ".",
                    "probability": 0.95458984375
                },
                {
                    "start": 1004.35,
                    "end": 1004.83,
                    "word": " In",
                    "probability": 0.80224609375
                },
                {
                    "start": 1004.83,
                    "end": 1005.05,
                    "word": " this",
                    "probability": 0.88232421875
                },
                {
                    "start": 1005.05,
                    "end": 1005.29,
                    "word": " demo",
                    "probability": 0.76171875
                },
                {
                    "start": 1005.29,
                    "end": 1005.41,
                    "word": ",",
                    "probability": 0.5712890625
                },
                {
                    "start": 1005.41,
                    "end": 1005.55,
                    "word": " we're",
                    "probability": 0.884765625
                },
                {
                    "start": 1005.55,
                    "end": 1005.65,
                    "word": " going",
                    "probability": 0.77734375
                },
                {
                    "start": 1005.65,
                    "end": 1005.79,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 1005.79,
                    "end": 1006.17,
                    "word": " show",
                    "probability": 0.861328125
                },
                {
                    "start": 1006.17,
                    "end": 1007.11,
                    "word": " how",
                    "probability": 0.8466796875
                },
                {
                    "start": 1007.11,
                    "end": 1007.31,
                    "word": " you're",
                    "probability": 0.5572509765625
                },
                {
                    "start": 1007.31,
                    "end": 1007.37,
                    "word": " going",
                    "probability": 0.833984375
                },
                {
                    "start": 1007.37,
                    "end": 1007.49,
                    "word": " to",
                    "probability": 0.8837890625
                },
                {
                    "start": 1007.49,
                    "end": 1007.65,
                    "word": " implement",
                    "probability": 0.841796875
                },
                {
                    "start": 1007.79,
                    "end": 1008.03,
                    "word": " a solution",
                    "probability": 0.8427734375
                },
                {
                    "start": 1008.03,
                    "end": 1008.27,
                    "word": " from",
                    "probability": 0.81494140625
                }
            ]
        },
        {
            "id": 153,
            "seek": 100830,
            "start": 1008.27,
            "end": 1012.61,
            "text": " from an adversary point of view, and then from a defense point of view, how you can easily apply",
            "tokens": [
                51520,
                490,
                364,
                48222,
                935,
                295,
                1910,
                11,
                293,
                550,
                490,
                257,
                7654,
                935,
                295,
                1910,
                11,
                577,
                291,
                393,
                3612,
                3079,
                51744
            ],
            "temperature": 0.0,
            "avg_logprob": -0.22323802541545096,
            "compression_ratio": 1.6977777777777778,
            "no_speech_prob": 0.010986328125,
            "words": [
                {
                    "start": 1008.27,
                    "end": 1008.39,
                    "word": " an",
                    "probability": 0.88671875
                },
                {
                    "start": 1008.39,
                    "end": 1008.73,
                    "word": " adversary",
                    "probability": 0.43408203125
                },
                {
                    "start": 1008.73,
                    "end": 1008.99,
                    "word": " point",
                    "probability": 0.84765625
                },
                {
                    "start": 1008.99,
                    "end": 1009.13,
                    "word": " of",
                    "probability": 0.89892578125
                },
                {
                    "start": 1009.13,
                    "end": 1009.29,
                    "word": " view",
                    "probability": 0.80078125
                },
                {
                    "start": 1009.29,
                    "end": 1009.41,
                    "word": ",",
                    "probability": 0.27734375
                },
                {
                    "start": 1009.41,
                    "end": 1009.59,
                    "word": " and",
                    "probability": 0.8896484375
                },
                {
                    "start": 1009.59,
                    "end": 1009.75,
                    "word": " then",
                    "probability": 0.7177734375
                },
                {
                    "start": 1009.75,
                    "end": 1009.93,
                    "word": " from",
                    "probability": 0.802734375
                },
                {
                    "start": 1010.03,
                    "end": 1010.33,
                    "word": " a defense",
                    "probability": 0.8310546875
                },
                {
                    "start": 1010.33,
                    "end": 1010.53,
                    "word": " point",
                    "probability": 0.8544921875
                },
                {
                    "start": 1010.53,
                    "end": 1010.65,
                    "word": " of",
                    "probability": 0.90283203125
                },
                {
                    "start": 1010.65,
                    "end": 1010.87,
                    "word": " view",
                    "probability": 0.81494140625
                },
                {
                    "start": 1010.87,
                    "end": 1011.31,
                    "word": ",",
                    "probability": 0.7939453125
                },
                {
                    "start": 1011.31,
                    "end": 1011.47,
                    "word": " how",
                    "probability": 0.85888671875
                },
                {
                    "start": 1011.47,
                    "end": 1011.53,
                    "word": " you",
                    "probability": 0.8701171875
                },
                {
                    "start": 1011.53,
                    "end": 1011.69,
                    "word": " can",
                    "probability": 0.79345703125
                },
                {
                    "start": 1011.69,
                    "end": 1012.25,
                    "word": " easily",
                    "probability": 0.845703125
                },
                {
                    "start": 1012.25,
                    "end": 1012.61,
                    "word": " apply",
                    "probability": 0.828125
                }
            ]
        },
        {
            "id": 154,
            "seek": 103212,
            "start": 1012.61,
            "end": 1019.35,
            "text": " differentiable private method into your model. So here in the references, you'll see the data",
            "tokens": [
                50364,
                819,
                9364,
                4551,
                3170,
                666,
                428,
                2316,
                13,
                407,
                510,
                294,
                264,
                15400,
                11,
                291,
                603,
                536,
                264,
                1412,
                50700
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3132972005678683,
            "compression_ratio": 1.641025641025641,
            "no_speech_prob": 0.00033545494079589844,
            "words": [
                {
                    "start": 1012.61,
                    "end": 1013.25,
                    "word": " differentiable",
                    "probability": 0.475921630859375
                },
                {
                    "start": 1013.25,
                    "end": 1013.51,
                    "word": " private",
                    "probability": 0.7861328125
                },
                {
                    "start": 1013.51,
                    "end": 1013.97,
                    "word": " method",
                    "probability": 0.88037109375
                },
                {
                    "start": 1013.97,
                    "end": 1014.69,
                    "word": " into",
                    "probability": 0.71533203125
                },
                {
                    "start": 1014.69,
                    "end": 1014.87,
                    "word": " your",
                    "probability": 0.810546875
                },
                {
                    "start": 1014.87,
                    "end": 1015.13,
                    "word": " model",
                    "probability": 0.8046875
                },
                {
                    "start": 1015.13,
                    "end": 1015.57,
                    "word": ".",
                    "probability": 0.931640625
                },
                {
                    "start": 1016.25,
                    "end": 1016.35,
                    "word": " So",
                    "probability": 0.54150390625
                },
                {
                    "start": 1016.35,
                    "end": 1016.81,
                    "word": " here",
                    "probability": 0.654296875
                },
                {
                    "start": 1016.81,
                    "end": 1017.01,
                    "word": " in",
                    "probability": 0.849609375
                },
                {
                    "start": 1017.01,
                    "end": 1017.19,
                    "word": " the",
                    "probability": 0.233154296875
                },
                {
                    "start": 1017.19,
                    "end": 1017.87,
                    "word": " references",
                    "probability": 0.81396484375
                },
                {
                    "start": 1017.87,
                    "end": 1018.27,
                    "word": ",",
                    "probability": 0.658203125
                },
                {
                    "start": 1018.27,
                    "end": 1018.77,
                    "word": " you'll",
                    "probability": 0.6729736328125
                },
                {
                    "start": 1018.77,
                    "end": 1018.99,
                    "word": " see",
                    "probability": 0.81982421875
                },
                {
                    "start": 1018.99,
                    "end": 1019.17,
                    "word": " the",
                    "probability": 0.81005859375
                },
                {
                    "start": 1019.17,
                    "end": 1019.35,
                    "word": " data",
                    "probability": 0.82421875
                }
            ]
        },
        {
            "id": 155,
            "seek": 103212,
            "start": 1019.35,
            "end": 1025.05,
            "text": " that we use, which is a data source on Kaggle of the purchase dataset. This contains information",
            "tokens": [
                50700,
                300,
                321,
                764,
                11,
                597,
                307,
                257,
                1412,
                4009,
                322,
                48751,
                22631,
                295,
                264,
                8110,
                28872,
                13,
                639,
                8306,
                1589,
                50988
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3132972005678683,
            "compression_ratio": 1.641025641025641,
            "no_speech_prob": 0.00033545494079589844,
            "words": [
                {
                    "start": 1019.35,
                    "end": 1019.49,
                    "word": " that",
                    "probability": 0.734375
                },
                {
                    "start": 1019.49,
                    "end": 1019.57,
                    "word": " we",
                    "probability": 0.85986328125
                },
                {
                    "start": 1019.57,
                    "end": 1019.89,
                    "word": " use",
                    "probability": 0.62353515625
                },
                {
                    "start": 1019.89,
                    "end": 1020.01,
                    "word": ",",
                    "probability": 0.802734375
                },
                {
                    "start": 1020.01,
                    "end": 1020.11,
                    "word": " which",
                    "probability": 0.9267578125
                },
                {
                    "start": 1020.11,
                    "end": 1020.41,
                    "word": " is",
                    "probability": 0.9052734375
                },
                {
                    "start": 1020.65,
                    "end": 1020.75,
                    "word": " a data",
                    "probability": 0.8193359375
                },
                {
                    "start": 1020.75,
                    "end": 1020.97,
                    "word": " source",
                    "probability": 0.6865234375
                },
                {
                    "start": 1020.97,
                    "end": 1021.17,
                    "word": " on",
                    "probability": 0.73876953125
                },
                {
                    "start": 1021.17,
                    "end": 1021.51,
                    "word": " Kaggle",
                    "probability": 0.92724609375
                },
                {
                    "start": 1021.51,
                    "end": 1021.83,
                    "word": " of",
                    "probability": 0.83837890625
                },
                {
                    "start": 1021.83,
                    "end": 1021.97,
                    "word": " the",
                    "probability": 0.8173828125
                },
                {
                    "start": 1021.97,
                    "end": 1022.19,
                    "word": " purchase",
                    "probability": 0.473876953125
                },
                {
                    "start": 1022.19,
                    "end": 1022.73,
                    "word": " dataset",
                    "probability": 0.486328125
                },
                {
                    "start": 1022.73,
                    "end": 1023.17,
                    "word": ".",
                    "probability": 0.947265625
                },
                {
                    "start": 1023.51,
                    "end": 1023.95,
                    "word": " This",
                    "probability": 0.80615234375
                },
                {
                    "start": 1023.95,
                    "end": 1024.39,
                    "word": " contains",
                    "probability": 0.8466796875
                },
                {
                    "start": 1024.39,
                    "end": 1025.05,
                    "word": " information",
                    "probability": 0.740234375
                }
            ]
        },
        {
            "id": 156,
            "seek": 103212,
            "start": 1025.67,
            "end": 1031.33,
            "text": " that you can use to create labels, such as a binary classifier. In some cases, it could be like",
            "tokens": [
                51044,
                300,
                291,
                393,
                764,
                281,
                1884,
                16949,
                11,
                1270,
                382,
                257,
                17434,
                1508,
                9902,
                13,
                682,
                512,
                3331,
                11,
                309,
                727,
                312,
                411,
                51300
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3132972005678683,
            "compression_ratio": 1.641025641025641,
            "no_speech_prob": 0.00033545494079589844,
            "words": [
                {
                    "start": 1025.67,
                    "end": 1026.11,
                    "word": " that",
                    "probability": 0.6416015625
                },
                {
                    "start": 1026.11,
                    "end": 1026.55,
                    "word": " you",
                    "probability": 0.8779296875
                },
                {
                    "start": 1026.55,
                    "end": 1026.69,
                    "word": " can",
                    "probability": 0.6484375
                },
                {
                    "start": 1026.69,
                    "end": 1027.11,
                    "word": " use",
                    "probability": 0.8076171875
                },
                {
                    "start": 1027.11,
                    "end": 1027.31,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 1027.31,
                    "end": 1027.71,
                    "word": " create",
                    "probability": 0.8515625
                },
                {
                    "start": 1027.71,
                    "end": 1028.05,
                    "word": " labels",
                    "probability": 0.859375
                },
                {
                    "start": 1028.05,
                    "end": 1028.23,
                    "word": ",",
                    "probability": 0.27294921875
                },
                {
                    "start": 1028.23,
                    "end": 1028.55,
                    "word": " such",
                    "probability": 0.8759765625
                },
                {
                    "start": 1028.55,
                    "end": 1028.69,
                    "word": " as",
                    "probability": 0.9013671875
                },
                {
                    "start": 1028.79,
                    "end": 1028.99,
                    "word": " a binary",
                    "probability": 0.7607421875
                },
                {
                    "start": 1028.99,
                    "end": 1029.57,
                    "word": " classifier",
                    "probability": 0.966796875
                },
                {
                    "start": 1029.57,
                    "end": 1030.01,
                    "word": ".",
                    "probability": 0.77978515625
                },
                {
                    "start": 1030.13,
                    "end": 1030.35,
                    "word": " In",
                    "probability": 0.87890625
                },
                {
                    "start": 1030.35,
                    "end": 1030.69,
                    "word": " some",
                    "probability": 0.810546875
                },
                {
                    "start": 1030.69,
                    "end": 1030.87,
                    "word": " cases",
                    "probability": 0.78515625
                },
                {
                    "start": 1030.87,
                    "end": 1030.89,
                    "word": ",",
                    "probability": 0.67578125
                },
                {
                    "start": 1030.89,
                    "end": 1030.97,
                    "word": " it",
                    "probability": 0.89404296875
                },
                {
                    "start": 1030.97,
                    "end": 1031.03,
                    "word": " could",
                    "probability": 0.755859375
                },
                {
                    "start": 1031.03,
                    "end": 1031.17,
                    "word": " be",
                    "probability": 0.89599609375
                },
                {
                    "start": 1031.17,
                    "end": 1031.33,
                    "word": " like",
                    "probability": 0.51904296875
                }
            ]
        },
        {
            "id": 157,
            "seek": 103212,
            "start": 1031.33,
            "end": 1036.43,
            "text": " fraud, or if it's fraudulent or not, or it could be abusive or not. Then we can also create a set",
            "tokens": [
                51300,
                14560,
                11,
                420,
                498,
                309,
                311,
                14560,
                23405,
                420,
                406,
                11,
                420,
                309,
                727,
                312,
                32828,
                420,
                406,
                13,
                1396,
                321,
                393,
                611,
                1884,
                257,
                992,
                51552
            ],
            "temperature": 0.0,
            "avg_logprob": -0.3132972005678683,
            "compression_ratio": 1.641025641025641,
            "no_speech_prob": 0.00033545494079589844,
            "words": [
                {
                    "start": 1031.33,
                    "end": 1031.63,
                    "word": " fraud",
                    "probability": 0.0357666015625
                },
                {
                    "start": 1031.63,
                    "end": 1031.79,
                    "word": ",",
                    "probability": 0.367919921875
                },
                {
                    "start": 1031.79,
                    "end": 1031.79,
                    "word": " or",
                    "probability": 0.599609375
                },
                {
                    "start": 1031.79,
                    "end": 1032.43,
                    "word": " if",
                    "probability": 0.86181640625
                },
                {
                    "start": 1032.43,
                    "end": 1032.63,
                    "word": " it's",
                    "probability": 0.948974609375
                },
                {
                    "start": 1032.63,
                    "end": 1032.89,
                    "word": " fraudulent",
                    "probability": 0.92626953125
                },
                {
                    "start": 1032.89,
                    "end": 1033.05,
                    "word": " or",
                    "probability": 0.87255859375
                },
                {
                    "start": 1033.05,
                    "end": 1033.33,
                    "word": " not",
                    "probability": 0.896484375
                },
                {
                    "start": 1033.33,
                    "end": 1033.39,
                    "word": ",",
                    "probability": 0.9072265625
                },
                {
                    "start": 1033.39,
                    "end": 1033.59,
                    "word": " or",
                    "probability": 0.8916015625
                },
                {
                    "start": 1033.59,
                    "end": 1033.69,
                    "word": " it",
                    "probability": 0.81982421875
                },
                {
                    "start": 1033.69,
                    "end": 1033.73,
                    "word": " could",
                    "probability": 0.81103515625
                },
                {
                    "start": 1033.73,
                    "end": 1033.85,
                    "word": " be",
                    "probability": 0.89111328125
                },
                {
                    "start": 1033.85,
                    "end": 1034.13,
                    "word": " abusive",
                    "probability": 0.82080078125
                },
                {
                    "start": 1034.13,
                    "end": 1034.31,
                    "word": " or",
                    "probability": 0.87451171875
                },
                {
                    "start": 1034.31,
                    "end": 1034.55,
                    "word": " not",
                    "probability": 0.89990234375
                },
                {
                    "start": 1034.55,
                    "end": 1034.91,
                    "word": ".",
                    "probability": 0.8603515625
                },
                {
                    "start": 1034.91,
                    "end": 1035.07,
                    "word": " Then",
                    "probability": 0.625
                },
                {
                    "start": 1035.07,
                    "end": 1035.23,
                    "word": " we",
                    "probability": 0.5546875
                },
                {
                    "start": 1035.23,
                    "end": 1035.39,
                    "word": " can",
                    "probability": 0.59521484375
                },
                {
                    "start": 1035.39,
                    "end": 1035.59,
                    "word": " also",
                    "probability": 0.8271484375
                },
                {
                    "start": 1035.59,
                    "end": 1036.05,
                    "word": " create",
                    "probability": 0.8310546875
                },
                {
                    "start": 1036.27,
                    "end": 1036.43,
                    "word": " a set",
                    "probability": 0.8427734375
                }
            ]
        },
        {
            "id": 158,
            "seek": 105984,
            "start": 1036.43,
            "end": 1045.73,
            "text": " of multi-class labels itself, such as the class for a customer, or it could be the category of a",
            "tokens": [
                50364,
                295,
                4825,
                12,
                11665,
                16949,
                2564,
                11,
                1270,
                382,
                264,
                1508,
                337,
                257,
                5474,
                11,
                420,
                309,
                727,
                312,
                264,
                7719,
                295,
                257,
                50812
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2592821817586918,
            "compression_ratio": 1.6695278969957081,
            "no_speech_prob": 0.0021820068359375,
            "words": [
                {
                    "start": 1036.43,
                    "end": 1036.45,
                    "word": " of",
                    "probability": 0.1451416015625
                },
                {
                    "start": 1036.45,
                    "end": 1036.77,
                    "word": " multi",
                    "probability": 0.11669921875
                },
                {
                    "start": 1036.77,
                    "end": 1037.17,
                    "word": "-class",
                    "probability": 0.87353515625
                },
                {
                    "start": 1037.17,
                    "end": 1037.45,
                    "word": " labels",
                    "probability": 0.192138671875
                },
                {
                    "start": 1037.45,
                    "end": 1038.11,
                    "word": " itself",
                    "probability": 0.65185546875
                },
                {
                    "start": 1038.11,
                    "end": 1038.37,
                    "word": ",",
                    "probability": 0.70703125
                },
                {
                    "start": 1038.37,
                    "end": 1039.11,
                    "word": " such",
                    "probability": 0.87841796875
                },
                {
                    "start": 1039.11,
                    "end": 1039.93,
                    "word": " as",
                    "probability": 0.89990234375
                },
                {
                    "start": 1039.93,
                    "end": 1040.33,
                    "word": " the",
                    "probability": 0.78076171875
                },
                {
                    "start": 1040.33,
                    "end": 1041.23,
                    "word": " class",
                    "probability": 0.9208984375
                },
                {
                    "start": 1041.23,
                    "end": 1041.61,
                    "word": " for",
                    "probability": 0.88330078125
                },
                {
                    "start": 1041.79,
                    "end": 1042.09,
                    "word": " a customer",
                    "probability": 0.6904296875
                },
                {
                    "start": 1042.09,
                    "end": 1042.49,
                    "word": ",",
                    "probability": 0.4453125
                },
                {
                    "start": 1042.49,
                    "end": 1042.57,
                    "word": " or",
                    "probability": 0.892578125
                },
                {
                    "start": 1042.57,
                    "end": 1042.71,
                    "word": " it",
                    "probability": 0.78466796875
                },
                {
                    "start": 1042.71,
                    "end": 1042.81,
                    "word": " could",
                    "probability": 0.81298828125
                },
                {
                    "start": 1042.81,
                    "end": 1043.57,
                    "word": " be",
                    "probability": 0.904296875
                },
                {
                    "start": 1043.57,
                    "end": 1044.39,
                    "word": " the",
                    "probability": 0.55224609375
                },
                {
                    "start": 1044.39,
                    "end": 1045.11,
                    "word": " category",
                    "probability": 0.83056640625
                },
                {
                    "start": 1045.11,
                    "end": 1045.35,
                    "word": " of",
                    "probability": 0.8046875
                },
                {
                    "start": 1045.45,
                    "end": 1045.73,
                    "word": " a merchant",
                    "probability": 0.6123046875
                }
            ]
        },
        {
            "id": 159,
            "seek": 105984,
            "start": 1045.73,
            "end": 1052.69,
            "text": " merchant, or any other multi-class that you can think of relevant to your use case. And you can",
            "tokens": [
                50812,
                32267,
                11,
                420,
                604,
                661,
                4825,
                12,
                11665,
                300,
                291,
                393,
                519,
                295,
                7340,
                281,
                428,
                764,
                1389,
                13,
                400,
                291,
                393,
                51164
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2592821817586918,
            "compression_ratio": 1.6695278969957081,
            "no_speech_prob": 0.0021820068359375,
            "words": [
                {
                    "start": 1045.73,
                    "end": 1046.27,
                    "word": ",",
                    "probability": 0.78466796875
                },
                {
                    "start": 1046.27,
                    "end": 1047.15,
                    "word": " or",
                    "probability": 0.90283203125
                },
                {
                    "start": 1047.15,
                    "end": 1047.41,
                    "word": " any",
                    "probability": 0.8203125
                },
                {
                    "start": 1047.41,
                    "end": 1047.77,
                    "word": " other",
                    "probability": 0.7783203125
                },
                {
                    "start": 1047.77,
                    "end": 1048.13,
                    "word": " multi",
                    "probability": 0.57666015625
                },
                {
                    "start": 1048.13,
                    "end": 1048.33,
                    "word": "-class",
                    "probability": 0.906982421875
                },
                {
                    "start": 1048.33,
                    "end": 1048.47,
                    "word": " that",
                    "probability": 0.83349609375
                },
                {
                    "start": 1048.47,
                    "end": 1048.55,
                    "word": " you",
                    "probability": 0.880859375
                },
                {
                    "start": 1048.55,
                    "end": 1048.63,
                    "word": " can",
                    "probability": 0.80322265625
                },
                {
                    "start": 1048.63,
                    "end": 1048.83,
                    "word": " think",
                    "probability": 0.80078125
                },
                {
                    "start": 1048.83,
                    "end": 1049.07,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 1049.07,
                    "end": 1049.49,
                    "word": " relevant",
                    "probability": 0.8271484375
                },
                {
                    "start": 1049.49,
                    "end": 1049.77,
                    "word": " to",
                    "probability": 0.90185546875
                },
                {
                    "start": 1049.77,
                    "end": 1049.91,
                    "word": " your",
                    "probability": 0.80908203125
                },
                {
                    "start": 1049.91,
                    "end": 1050.13,
                    "word": " use",
                    "probability": 0.77783203125
                },
                {
                    "start": 1050.13,
                    "end": 1050.43,
                    "word": " case",
                    "probability": 0.80712890625
                },
                {
                    "start": 1050.43,
                    "end": 1050.95,
                    "word": ".",
                    "probability": 0.982421875
                },
                {
                    "start": 1051.97,
                    "end": 1052.29,
                    "word": " And",
                    "probability": 0.278076171875
                },
                {
                    "start": 1052.29,
                    "end": 1052.39,
                    "word": " you",
                    "probability": 0.88623046875
                },
                {
                    "start": 1052.39,
                    "end": 1052.49,
                    "word": " can",
                    "probability": 0.7099609375
                },
                {
                    "start": 1052.49,
                    "end": 1052.69,
                    "word": " see",
                    "probability": 0.82080078125
                }
            ]
        },
        {
            "id": 160,
            "seek": 105984,
            "start": 1052.69,
            "end": 1057.85,
            "text": " see that we're also going to show you the TensorFlow privacy method as well. So first, we're going to",
            "tokens": [
                51164,
                536,
                300,
                321,
                434,
                611,
                516,
                281,
                855,
                291,
                264,
                37624,
                11427,
                3170,
                382,
                731,
                13,
                407,
                700,
                11,
                321,
                434,
                516,
                281,
                51424
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2592821817586918,
            "compression_ratio": 1.6695278969957081,
            "no_speech_prob": 0.0021820068359375,
            "words": [
                {
                    "start": 1052.69,
                    "end": 1052.91,
                    "word": " that",
                    "probability": 0.8837890625
                },
                {
                    "start": 1052.91,
                    "end": 1053.13,
                    "word": " we're",
                    "probability": 0.917724609375
                },
                {
                    "start": 1053.13,
                    "end": 1053.35,
                    "word": " also",
                    "probability": 0.810546875
                },
                {
                    "start": 1053.35,
                    "end": 1053.49,
                    "word": " going",
                    "probability": 0.7900390625
                },
                {
                    "start": 1053.49,
                    "end": 1053.59,
                    "word": " to",
                    "probability": 0.908203125
                },
                {
                    "start": 1053.59,
                    "end": 1053.75,
                    "word": " show",
                    "probability": 0.8779296875
                },
                {
                    "start": 1053.75,
                    "end": 1053.89,
                    "word": " you",
                    "probability": 0.8974609375
                },
                {
                    "start": 1053.89,
                    "end": 1054.07,
                    "word": " the",
                    "probability": 0.8154296875
                },
                {
                    "start": 1054.07,
                    "end": 1054.35,
                    "word": " TensorFlow",
                    "probability": 0.332763671875
                },
                {
                    "start": 1054.35,
                    "end": 1054.75,
                    "word": " privacy",
                    "probability": 0.61376953125
                },
                {
                    "start": 1054.75,
                    "end": 1055.25,
                    "word": " method",
                    "probability": 0.88134765625
                },
                {
                    "start": 1055.25,
                    "end": 1055.81,
                    "word": " as",
                    "probability": 0.84033203125
                },
                {
                    "start": 1055.81,
                    "end": 1056.03,
                    "word": " well",
                    "probability": 0.87744140625
                },
                {
                    "start": 1056.03,
                    "end": 1056.55,
                    "word": ".",
                    "probability": 0.99853515625
                },
                {
                    "start": 1056.95,
                    "end": 1057.09,
                    "word": " So",
                    "probability": 0.72607421875
                },
                {
                    "start": 1057.09,
                    "end": 1057.37,
                    "word": " first",
                    "probability": 0.7080078125
                },
                {
                    "start": 1057.37,
                    "end": 1057.43,
                    "word": ",",
                    "probability": 0.61328125
                },
                {
                    "start": 1057.43,
                    "end": 1057.55,
                    "word": " we're",
                    "probability": 0.929931640625
                },
                {
                    "start": 1057.55,
                    "end": 1057.59,
                    "word": " going",
                    "probability": 0.84619140625
                },
                {
                    "start": 1057.59,
                    "end": 1057.67,
                    "word": " to",
                    "probability": 0.9033203125
                },
                {
                    "start": 1057.67,
                    "end": 1057.85,
                    "word": " look",
                    "probability": 0.91943359375
                }
            ]
        },
        {
            "id": 161,
            "seek": 105984,
            "start": 1057.85,
            "end": 1064.15,
            "text": " look at how a shadow network is created. It's really easy. First, you create the set of shadow",
            "tokens": [
                51424,
                574,
                412,
                577,
                257,
                402,
                11528,
                3209,
                307,
                2942,
                13,
                467,
                311,
                534,
                1858,
                13,
                2386,
                11,
                291,
                1884,
                264,
                992,
                295,
                8576,
                51752
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2592821817586918,
            "compression_ratio": 1.6695278969957081,
            "no_speech_prob": 0.0021820068359375,
            "words": [
                {
                    "start": 1057.85,
                    "end": 1058.23,
                    "word": " at",
                    "probability": 0.91064453125
                },
                {
                    "start": 1058.23,
                    "end": 1059.15,
                    "word": " how",
                    "probability": 0.8974609375
                },
                {
                    "start": 1059.39,
                    "end": 1059.51,
                    "word": " a shadow",
                    "probability": 0.4923095703125
                },
                {
                    "start": 1059.51,
                    "end": 1059.81,
                    "word": " network",
                    "probability": 0.8310546875
                },
                {
                    "start": 1059.81,
                    "end": 1059.99,
                    "word": " is",
                    "probability": 0.90625
                },
                {
                    "start": 1059.99,
                    "end": 1060.33,
                    "word": " created",
                    "probability": 0.73486328125
                },
                {
                    "start": 1060.33,
                    "end": 1060.49,
                    "word": ".",
                    "probability": 0.94677734375
                },
                {
                    "start": 1060.49,
                    "end": 1060.79,
                    "word": " It's",
                    "probability": 0.95361328125
                },
                {
                    "start": 1060.79,
                    "end": 1061.17,
                    "word": " really",
                    "probability": 0.7841796875
                },
                {
                    "start": 1061.17,
                    "end": 1061.53,
                    "word": " easy",
                    "probability": 0.810546875
                },
                {
                    "start": 1061.53,
                    "end": 1061.85,
                    "word": ".",
                    "probability": 0.96826171875
                },
                {
                    "start": 1061.85,
                    "end": 1062.11,
                    "word": " First",
                    "probability": 0.876953125
                },
                {
                    "start": 1062.11,
                    "end": 1062.21,
                    "word": ",",
                    "probability": 0.85205078125
                },
                {
                    "start": 1062.21,
                    "end": 1062.31,
                    "word": " you",
                    "probability": 0.875
                },
                {
                    "start": 1062.31,
                    "end": 1062.73,
                    "word": " create",
                    "probability": 0.837890625
                },
                {
                    "start": 1062.73,
                    "end": 1063.15,
                    "word": " the",
                    "probability": 0.70068359375
                },
                {
                    "start": 1063.15,
                    "end": 1063.79,
                    "word": " set",
                    "probability": 0.86083984375
                },
                {
                    "start": 1063.79,
                    "end": 1063.93,
                    "word": " of",
                    "probability": 0.89990234375
                },
                {
                    "start": 1063.93,
                    "end": 1064.15,
                    "word": " shadow",
                    "probability": 0.57568359375
                }
            ]
        },
        {
            "id": 162,
            "seek": 108984,
            "start": 1064.15,
            "end": 1070.45,
            "text": " networks. Then you output the predictions from the shadow networks, feed that into a TAC network,",
            "tokens": [
                50364,
                9590,
                13,
                1396,
                291,
                5598,
                264,
                21264,
                490,
                264,
                8576,
                9590,
                11,
                3154,
                300,
                666,
                257,
                314,
                4378,
                3209,
                11,
                50684
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2532327573361068,
            "compression_ratio": 1.9715447154471544,
            "no_speech_prob": 0.0004107952117919922,
            "words": [
                {
                    "start": 1064.15,
                    "end": 1064.57,
                    "word": " networks",
                    "probability": 0.291015625
                },
                {
                    "start": 1064.57,
                    "end": 1064.79,
                    "word": ".",
                    "probability": 0.1907958984375
                },
                {
                    "start": 1064.85,
                    "end": 1065.33,
                    "word": " Then",
                    "probability": 0.78857421875
                },
                {
                    "start": 1065.33,
                    "end": 1065.73,
                    "word": " you",
                    "probability": 0.841796875
                },
                {
                    "start": 1065.73,
                    "end": 1066.37,
                    "word": " output",
                    "probability": 0.59912109375
                },
                {
                    "start": 1066.37,
                    "end": 1066.81,
                    "word": " the",
                    "probability": 0.77490234375
                },
                {
                    "start": 1066.81,
                    "end": 1067.17,
                    "word": " predictions",
                    "probability": 0.67724609375
                },
                {
                    "start": 1067.17,
                    "end": 1067.43,
                    "word": " from",
                    "probability": 0.7451171875
                },
                {
                    "start": 1067.43,
                    "end": 1067.55,
                    "word": " the",
                    "probability": 0.802734375
                },
                {
                    "start": 1067.55,
                    "end": 1067.73,
                    "word": " shadow",
                    "probability": 0.1871337890625
                },
                {
                    "start": 1067.73,
                    "end": 1068.23,
                    "word": " networks",
                    "probability": 0.6328125
                },
                {
                    "start": 1068.23,
                    "end": 1068.47,
                    "word": ",",
                    "probability": 0.7861328125
                },
                {
                    "start": 1068.47,
                    "end": 1068.63,
                    "word": " feed",
                    "probability": 0.85986328125
                },
                {
                    "start": 1068.63,
                    "end": 1068.83,
                    "word": " that",
                    "probability": 0.87255859375
                },
                {
                    "start": 1068.83,
                    "end": 1069.13,
                    "word": " into",
                    "probability": 0.7294921875
                },
                {
                    "start": 1069.35,
                    "end": 1069.61,
                    "word": " a TAC",
                    "probability": 0.583984375
                },
                {
                    "start": 1069.61,
                    "end": 1069.95,
                    "word": " network",
                    "probability": 0.8134765625
                },
                {
                    "start": 1069.95,
                    "end": 1070.45,
                    "word": ",",
                    "probability": 0.64404296875
                }
            ]
        },
        {
            "id": 163,
            "seek": 108984,
            "start": 1070.45,
            "end": 1075.91,
            "text": " and then train a target network that is similar to what you think the production environment would",
            "tokens": [
                50684,
                293,
                550,
                3847,
                257,
                3779,
                3209,
                300,
                307,
                2531,
                281,
                437,
                291,
                519,
                264,
                4265,
                2823,
                576,
                50952
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2532327573361068,
            "compression_ratio": 1.9715447154471544,
            "no_speech_prob": 0.0004107952117919922,
            "words": [
                {
                    "start": 1070.45,
                    "end": 1070.81,
                    "word": " and",
                    "probability": 0.8798828125
                },
                {
                    "start": 1070.81,
                    "end": 1071.19,
                    "word": " then",
                    "probability": 0.7412109375
                },
                {
                    "start": 1071.19,
                    "end": 1071.67,
                    "word": " train",
                    "probability": 0.7158203125
                },
                {
                    "start": 1072.27,
                    "end": 1072.47,
                    "word": " a target",
                    "probability": 0.83935546875
                },
                {
                    "start": 1072.47,
                    "end": 1072.67,
                    "word": " network",
                    "probability": 0.671875
                },
                {
                    "start": 1072.67,
                    "end": 1072.89,
                    "word": " that",
                    "probability": 0.85888671875
                },
                {
                    "start": 1072.89,
                    "end": 1073.77,
                    "word": " is",
                    "probability": 0.8828125
                },
                {
                    "start": 1073.77,
                    "end": 1074.01,
                    "word": " similar",
                    "probability": 0.9248046875
                },
                {
                    "start": 1074.01,
                    "end": 1074.29,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 1074.29,
                    "end": 1074.51,
                    "word": " what",
                    "probability": 0.88134765625
                },
                {
                    "start": 1074.51,
                    "end": 1074.63,
                    "word": " you",
                    "probability": 0.900390625
                },
                {
                    "start": 1074.63,
                    "end": 1074.79,
                    "word": " think",
                    "probability": 0.79736328125
                },
                {
                    "start": 1074.79,
                    "end": 1074.93,
                    "word": " the",
                    "probability": 0.80126953125
                },
                {
                    "start": 1074.93,
                    "end": 1075.17,
                    "word": " production",
                    "probability": 0.77880859375
                },
                {
                    "start": 1075.17,
                    "end": 1075.65,
                    "word": " environment",
                    "probability": 0.5751953125
                },
                {
                    "start": 1075.65,
                    "end": 1075.91,
                    "word": " would",
                    "probability": 0.64892578125
                }
            ]
        },
        {
            "id": 164,
            "seek": 108984,
            "start": 1075.91,
            "end": 1082.01,
            "text": " look like, and then see whether the prediction from that target network is predicted as an in-member",
            "tokens": [
                50952,
                574,
                411,
                11,
                293,
                550,
                536,
                1968,
                264,
                17630,
                490,
                300,
                3779,
                3209,
                307,
                19147,
                382,
                364,
                294,
                12,
                38249,
                51256
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2532327573361068,
            "compression_ratio": 1.9715447154471544,
            "no_speech_prob": 0.0004107952117919922,
            "words": [
                {
                    "start": 1075.91,
                    "end": 1076.11,
                    "word": " look",
                    "probability": 0.91552734375
                },
                {
                    "start": 1076.11,
                    "end": 1076.33,
                    "word": " like",
                    "probability": 0.8798828125
                },
                {
                    "start": 1076.33,
                    "end": 1076.53,
                    "word": ",",
                    "probability": 0.426025390625
                },
                {
                    "start": 1076.53,
                    "end": 1076.65,
                    "word": " and",
                    "probability": 0.8857421875
                },
                {
                    "start": 1076.65,
                    "end": 1076.93,
                    "word": " then",
                    "probability": 0.7197265625
                },
                {
                    "start": 1076.93,
                    "end": 1077.57,
                    "word": " see",
                    "probability": 0.8076171875
                },
                {
                    "start": 1077.57,
                    "end": 1078.07,
                    "word": " whether",
                    "probability": 0.84423828125
                },
                {
                    "start": 1078.07,
                    "end": 1078.81,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 1078.81,
                    "end": 1079.05,
                    "word": " prediction",
                    "probability": 0.8818359375
                },
                {
                    "start": 1079.05,
                    "end": 1079.23,
                    "word": " from",
                    "probability": 0.66796875
                },
                {
                    "start": 1079.23,
                    "end": 1079.33,
                    "word": " that",
                    "probability": 0.28125
                },
                {
                    "start": 1079.33,
                    "end": 1079.65,
                    "word": " target",
                    "probability": 0.8330078125
                },
                {
                    "start": 1079.65,
                    "end": 1080.01,
                    "word": " network",
                    "probability": 0.85546875
                },
                {
                    "start": 1080.01,
                    "end": 1080.75,
                    "word": " is",
                    "probability": 0.880859375
                },
                {
                    "start": 1080.75,
                    "end": 1081.11,
                    "word": " predicted",
                    "probability": 0.72216796875
                },
                {
                    "start": 1081.11,
                    "end": 1081.55,
                    "word": " as",
                    "probability": 0.9013671875
                },
                {
                    "start": 1081.55,
                    "end": 1081.71,
                    "word": " an",
                    "probability": 0.7216796875
                },
                {
                    "start": 1081.71,
                    "end": 1081.85,
                    "word": " in",
                    "probability": 0.90087890625
                },
                {
                    "start": 1081.85,
                    "end": 1082.01,
                    "word": "-member",
                    "probability": 0.69384765625
                }
            ]
        },
        {
            "id": 165,
            "seek": 108984,
            "start": 1082.01,
            "end": 1087.85,
            "text": " or out-member from the TAC network. If we look at the shadow network, all you have to do is create",
            "tokens": [
                51256,
                420,
                484,
                12,
                38249,
                490,
                264,
                314,
                4378,
                3209,
                13,
                759,
                321,
                574,
                412,
                264,
                8576,
                3209,
                11,
                439,
                291,
                362,
                281,
                360,
                307,
                1884,
                51552
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2532327573361068,
            "compression_ratio": 1.9715447154471544,
            "no_speech_prob": 0.0004107952117919922,
            "words": [
                {
                    "start": 1082.01,
                    "end": 1082.17,
                    "word": " or",
                    "probability": 0.85791015625
                },
                {
                    "start": 1082.17,
                    "end": 1082.33,
                    "word": " out",
                    "probability": 0.68505859375
                },
                {
                    "start": 1082.33,
                    "end": 1082.47,
                    "word": "-member",
                    "probability": 0.860595703125
                },
                {
                    "start": 1082.47,
                    "end": 1082.71,
                    "word": " from",
                    "probability": 0.71875
                },
                {
                    "start": 1082.71,
                    "end": 1082.85,
                    "word": " the",
                    "probability": 0.80224609375
                },
                {
                    "start": 1082.85,
                    "end": 1083.05,
                    "word": " TAC",
                    "probability": 0.923583984375
                },
                {
                    "start": 1083.05,
                    "end": 1083.41,
                    "word": " network",
                    "probability": 0.86279296875
                },
                {
                    "start": 1083.41,
                    "end": 1083.89,
                    "word": ".",
                    "probability": 0.994140625
                },
                {
                    "start": 1084.19,
                    "end": 1084.67,
                    "word": " If",
                    "probability": 0.884765625
                },
                {
                    "start": 1084.67,
                    "end": 1084.75,
                    "word": " we",
                    "probability": 0.67724609375
                },
                {
                    "start": 1084.75,
                    "end": 1084.89,
                    "word": " look",
                    "probability": 0.91650390625
                },
                {
                    "start": 1084.89,
                    "end": 1085.07,
                    "word": " at",
                    "probability": 0.912109375
                },
                {
                    "start": 1085.07,
                    "end": 1085.33,
                    "word": " the",
                    "probability": 0.81591796875
                },
                {
                    "start": 1085.33,
                    "end": 1085.53,
                    "word": " shadow",
                    "probability": 0.407958984375
                },
                {
                    "start": 1085.53,
                    "end": 1086.03,
                    "word": " network",
                    "probability": 0.81689453125
                },
                {
                    "start": 1086.03,
                    "end": 1086.39,
                    "word": ",",
                    "probability": 0.8828125
                },
                {
                    "start": 1086.39,
                    "end": 1086.77,
                    "word": " all",
                    "probability": 0.8466796875
                },
                {
                    "start": 1086.77,
                    "end": 1086.85,
                    "word": " you",
                    "probability": 0.88427734375
                },
                {
                    "start": 1086.85,
                    "end": 1086.93,
                    "word": " have",
                    "probability": 0.88134765625
                },
                {
                    "start": 1086.93,
                    "end": 1087.03,
                    "word": " to",
                    "probability": 0.9072265625
                },
                {
                    "start": 1087.03,
                    "end": 1087.25,
                    "word": " do",
                    "probability": 0.912109375
                },
                {
                    "start": 1087.25,
                    "end": 1087.49,
                    "word": " is",
                    "probability": 0.90087890625
                },
                {
                    "start": 1087.49,
                    "end": 1087.85,
                    "word": " create",
                    "probability": 0.84619140625
                }
            ]
        },
        {
            "id": 166,
            "seek": 108984,
            "start": 1088.13,
            "end": 1094.01,
            "text": " a very simple for loop. So if you go down, you can see that all you have to do is create",
            "tokens": [
                51552,
                257,
                588,
                2199,
                337,
                6367,
                13,
                407,
                498,
                291,
                352,
                760,
                11,
                291,
                393,
                536,
                300,
                439,
                291,
                362,
                281,
                360,
                307,
                1884,
                51856
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2532327573361068,
            "compression_ratio": 1.9715447154471544,
            "no_speech_prob": 0.0004107952117919922,
            "words": [
                {
                    "start": 1088.13,
                    "end": 1088.33,
                    "word": " a very",
                    "probability": 0.73193359375
                },
                {
                    "start": 1088.33,
                    "end": 1088.73,
                    "word": " simple",
                    "probability": 0.91455078125
                },
                {
                    "start": 1088.73,
                    "end": 1089.03,
                    "word": " for",
                    "probability": 0.81640625
                },
                {
                    "start": 1089.03,
                    "end": 1089.29,
                    "word": " loop",
                    "probability": 0.72900390625
                },
                {
                    "start": 1089.29,
                    "end": 1089.77,
                    "word": ".",
                    "probability": 0.83935546875
                },
                {
                    "start": 1089.83,
                    "end": 1089.91,
                    "word": " So",
                    "probability": 0.8359375
                },
                {
                    "start": 1089.91,
                    "end": 1090.01,
                    "word": " if",
                    "probability": 0.78076171875
                },
                {
                    "start": 1090.01,
                    "end": 1090.09,
                    "word": " you",
                    "probability": 0.888671875
                },
                {
                    "start": 1090.09,
                    "end": 1090.23,
                    "word": " go",
                    "probability": 0.892578125
                },
                {
                    "start": 1090.23,
                    "end": 1090.51,
                    "word": " down",
                    "probability": 0.80712890625
                },
                {
                    "start": 1090.51,
                    "end": 1092.09,
                    "word": ",",
                    "probability": 0.80859375
                },
                {
                    "start": 1092.09,
                    "end": 1092.19,
                    "word": " you",
                    "probability": 0.88037109375
                },
                {
                    "start": 1092.19,
                    "end": 1092.29,
                    "word": " can",
                    "probability": 0.787109375
                },
                {
                    "start": 1092.29,
                    "end": 1092.47,
                    "word": " see",
                    "probability": 0.8232421875
                },
                {
                    "start": 1092.47,
                    "end": 1092.73,
                    "word": " that",
                    "probability": 0.8828125
                },
                {
                    "start": 1092.73,
                    "end": 1093.05,
                    "word": " all",
                    "probability": 0.88671875
                },
                {
                    "start": 1093.05,
                    "end": 1093.11,
                    "word": " you",
                    "probability": 0.86865234375
                },
                {
                    "start": 1093.11,
                    "end": 1093.19,
                    "word": " have",
                    "probability": 0.880859375
                },
                {
                    "start": 1093.19,
                    "end": 1093.27,
                    "word": " to",
                    "probability": 0.90966796875
                },
                {
                    "start": 1093.27,
                    "end": 1093.45,
                    "word": " do",
                    "probability": 0.90966796875
                },
                {
                    "start": 1093.45,
                    "end": 1093.65,
                    "word": " is",
                    "probability": 0.900390625
                },
                {
                    "start": 1093.65,
                    "end": 1094.01,
                    "word": " create",
                    "probability": 0.85302734375
                }
            ]
        },
        {
            "id": 167,
            "seek": 111704,
            "start": 1094.15,
            "end": 1099.87,
            "text": " n number of shadow networks. They could be similar to each other, or they could be of different",
            "tokens": [
                50388,
                297,
                1230,
                295,
                8576,
                9590,
                13,
                814,
                727,
                312,
                2531,
                281,
                1184,
                661,
                11,
                420,
                436,
                727,
                312,
                295,
                819,
                50648
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2781894428213847,
            "compression_ratio": 1.6359832635983265,
            "no_speech_prob": 0.00044417381286621094,
            "words": [
                {
                    "start": 1094.15,
                    "end": 1094.61,
                    "word": " n",
                    "probability": 0.387939453125
                },
                {
                    "start": 1094.61,
                    "end": 1094.95,
                    "word": " number",
                    "probability": 0.80126953125
                },
                {
                    "start": 1094.95,
                    "end": 1095.53,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 1095.53,
                    "end": 1096.47,
                    "word": " shadow",
                    "probability": 0.319091796875
                },
                {
                    "start": 1096.47,
                    "end": 1097.03,
                    "word": " networks",
                    "probability": 0.61572265625
                },
                {
                    "start": 1097.03,
                    "end": 1097.19,
                    "word": ".",
                    "probability": 0.418701171875
                },
                {
                    "start": 1097.19,
                    "end": 1097.29,
                    "word": " They",
                    "probability": 0.72802734375
                },
                {
                    "start": 1097.29,
                    "end": 1097.43,
                    "word": " could",
                    "probability": 0.818359375
                },
                {
                    "start": 1097.43,
                    "end": 1097.61,
                    "word": " be",
                    "probability": 0.900390625
                },
                {
                    "start": 1097.61,
                    "end": 1097.81,
                    "word": " similar",
                    "probability": 0.92138671875
                },
                {
                    "start": 1097.81,
                    "end": 1098.01,
                    "word": " to",
                    "probability": 0.9091796875
                },
                {
                    "start": 1098.01,
                    "end": 1098.21,
                    "word": " each",
                    "probability": 0.85693359375
                },
                {
                    "start": 1098.21,
                    "end": 1098.51,
                    "word": " other",
                    "probability": 0.77978515625
                },
                {
                    "start": 1098.51,
                    "end": 1098.57,
                    "word": ",",
                    "probability": 0.2467041015625
                },
                {
                    "start": 1098.57,
                    "end": 1098.85,
                    "word": " or",
                    "probability": 0.9052734375
                },
                {
                    "start": 1098.85,
                    "end": 1099.05,
                    "word": " they",
                    "probability": 0.72021484375
                },
                {
                    "start": 1099.05,
                    "end": 1099.17,
                    "word": " could",
                    "probability": 0.8232421875
                },
                {
                    "start": 1099.17,
                    "end": 1099.43,
                    "word": " be",
                    "probability": 0.89306640625
                },
                {
                    "start": 1099.43,
                    "end": 1099.67,
                    "word": " of",
                    "probability": 0.83056640625
                },
                {
                    "start": 1099.67,
                    "end": 1099.87,
                    "word": " different",
                    "probability": 0.6845703125
                }
            ]
        },
        {
            "id": 168,
            "seek": 111704,
            "start": 1099.87,
            "end": 1106.69,
            "text": " structure. But for our use case, we're going to create one type of TensorFlow model, a very simple",
            "tokens": [
                50648,
                3877,
                13,
                583,
                337,
                527,
                764,
                1389,
                11,
                321,
                434,
                516,
                281,
                1884,
                472,
                2010,
                295,
                37624,
                2316,
                11,
                257,
                588,
                2199,
                50992
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2781894428213847,
            "compression_ratio": 1.6359832635983265,
            "no_speech_prob": 0.00044417381286621094,
            "words": [
                {
                    "start": 1099.87,
                    "end": 1100.23,
                    "word": " structure",
                    "probability": 0.69091796875
                },
                {
                    "start": 1100.23,
                    "end": 1100.33,
                    "word": ".",
                    "probability": 0.4599609375
                },
                {
                    "start": 1100.33,
                    "end": 1100.45,
                    "word": " But",
                    "probability": 0.873046875
                },
                {
                    "start": 1100.45,
                    "end": 1100.65,
                    "word": " for",
                    "probability": 0.8857421875
                },
                {
                    "start": 1100.65,
                    "end": 1100.89,
                    "word": " our",
                    "probability": 0.81494140625
                },
                {
                    "start": 1100.89,
                    "end": 1101.11,
                    "word": " use",
                    "probability": 0.814453125
                },
                {
                    "start": 1101.11,
                    "end": 1101.47,
                    "word": " case",
                    "probability": 0.79296875
                },
                {
                    "start": 1101.47,
                    "end": 1101.93,
                    "word": ",",
                    "probability": 0.93359375
                },
                {
                    "start": 1101.93,
                    "end": 1102.07,
                    "word": " we're",
                    "probability": 0.797607421875
                },
                {
                    "start": 1102.07,
                    "end": 1102.21,
                    "word": " going",
                    "probability": 0.5830078125
                },
                {
                    "start": 1102.21,
                    "end": 1102.31,
                    "word": " to",
                    "probability": 0.90185546875
                },
                {
                    "start": 1102.31,
                    "end": 1102.69,
                    "word": " create",
                    "probability": 0.83056640625
                },
                {
                    "start": 1102.69,
                    "end": 1103.29,
                    "word": " one",
                    "probability": 0.806640625
                },
                {
                    "start": 1103.29,
                    "end": 1103.71,
                    "word": " type",
                    "probability": 0.916015625
                },
                {
                    "start": 1103.71,
                    "end": 1103.97,
                    "word": " of",
                    "probability": 0.90283203125
                },
                {
                    "start": 1103.97,
                    "end": 1105.05,
                    "word": " TensorFlow",
                    "probability": 0.80126953125
                },
                {
                    "start": 1105.05,
                    "end": 1105.53,
                    "word": " model",
                    "probability": 0.82568359375
                },
                {
                    "start": 1105.53,
                    "end": 1106.01,
                    "word": ",",
                    "probability": 0.728515625
                },
                {
                    "start": 1106.17,
                    "end": 1106.37,
                    "word": " a very",
                    "probability": 0.7255859375
                },
                {
                    "start": 1106.37,
                    "end": 1106.69,
                    "word": " simple",
                    "probability": 0.9130859375
                }
            ]
        },
        {
            "id": 169,
            "seek": 111704,
            "start": 1106.69,
            "end": 1115.07,
            "text": " one with an initial density of 64 nodes, and then dropout, and then another dense layer of 24 nodes,",
            "tokens": [
                50992,
                472,
                365,
                364,
                5883,
                24505,
                507,
                295,
                12145,
                13891,
                11,
                293,
                550,
                3270,
                346,
                11,
                293,
                550,
                1071,
                18011,
                4583,
                295,
                4022,
                13891,
                11,
                51416
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2781894428213847,
            "compression_ratio": 1.6359832635983265,
            "no_speech_prob": 0.00044417381286621094,
            "words": [
                {
                    "start": 1106.69,
                    "end": 1107.03,
                    "word": " one",
                    "probability": 0.82470703125
                },
                {
                    "start": 1107.03,
                    "end": 1107.33,
                    "word": " with",
                    "probability": 0.71435546875
                },
                {
                    "start": 1107.33,
                    "end": 1108.49,
                    "word": " an",
                    "probability": 0.49609375
                },
                {
                    "start": 1108.49,
                    "end": 1108.71,
                    "word": " initial",
                    "probability": 0.85791015625
                },
                {
                    "start": 1108.71,
                    "end": 1108.97,
                    "word": " density",
                    "probability": 0.551513671875
                },
                {
                    "start": 1108.97,
                    "end": 1109.13,
                    "word": " of",
                    "probability": 0.8818359375
                },
                {
                    "start": 1109.13,
                    "end": 1109.67,
                    "word": " 64",
                    "probability": 0.89013671875
                },
                {
                    "start": 1109.67,
                    "end": 1111.21,
                    "word": " nodes",
                    "probability": 0.60888671875
                },
                {
                    "start": 1111.21,
                    "end": 1111.35,
                    "word": ",",
                    "probability": 0.6298828125
                },
                {
                    "start": 1111.35,
                    "end": 1111.59,
                    "word": " and",
                    "probability": 0.86767578125
                },
                {
                    "start": 1111.59,
                    "end": 1111.89,
                    "word": " then",
                    "probability": 0.7373046875
                },
                {
                    "start": 1111.89,
                    "end": 1112.53,
                    "word": " dropout",
                    "probability": 0.626708984375
                },
                {
                    "start": 1112.53,
                    "end": 1112.63,
                    "word": ",",
                    "probability": 0.463134765625
                },
                {
                    "start": 1112.63,
                    "end": 1112.71,
                    "word": " and",
                    "probability": 0.8740234375
                },
                {
                    "start": 1112.71,
                    "end": 1112.83,
                    "word": " then",
                    "probability": 0.58740234375
                },
                {
                    "start": 1112.83,
                    "end": 1113.03,
                    "word": " another",
                    "probability": 0.72119140625
                },
                {
                    "start": 1113.03,
                    "end": 1113.39,
                    "word": " dense",
                    "probability": 0.488525390625
                },
                {
                    "start": 1113.39,
                    "end": 1113.59,
                    "word": " layer",
                    "probability": 0.75048828125
                },
                {
                    "start": 1113.59,
                    "end": 1113.91,
                    "word": " of",
                    "probability": 0.908203125
                },
                {
                    "start": 1113.91,
                    "end": 1114.37,
                    "word": " 24",
                    "probability": 0.8935546875
                },
                {
                    "start": 1114.37,
                    "end": 1114.79,
                    "word": " nodes",
                    "probability": 0.8173828125
                },
                {
                    "start": 1114.79,
                    "end": 1115.07,
                    "word": ",",
                    "probability": 0.358154296875
                }
            ]
        },
        {
            "id": 170,
            "seek": 111704,
            "start": 1115.07,
            "end": 1121.35,
            "text": " and then finally the softmax for a prediction of the multi-class or binary class. Each of these",
            "tokens": [
                51416,
                293,
                550,
                2721,
                264,
                2787,
                41167,
                337,
                257,
                17630,
                295,
                264,
                4825,
                12,
                11665,
                420,
                17434,
                1508,
                13,
                6947,
                295,
                613,
                51724
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2781894428213847,
            "compression_ratio": 1.6359832635983265,
            "no_speech_prob": 0.00044417381286621094,
            "words": [
                {
                    "start": 1115.07,
                    "end": 1115.43,
                    "word": " and",
                    "probability": 0.88525390625
                },
                {
                    "start": 1115.43,
                    "end": 1115.71,
                    "word": " then",
                    "probability": 0.71875
                },
                {
                    "start": 1115.71,
                    "end": 1116.05,
                    "word": " finally",
                    "probability": 0.708984375
                },
                {
                    "start": 1116.05,
                    "end": 1116.29,
                    "word": " the",
                    "probability": 0.5390625
                },
                {
                    "start": 1116.29,
                    "end": 1116.89,
                    "word": " softmax",
                    "probability": 0.796875
                },
                {
                    "start": 1116.89,
                    "end": 1117.33,
                    "word": " for",
                    "probability": 0.60595703125
                },
                {
                    "start": 1117.45,
                    "end": 1117.73,
                    "word": " a prediction",
                    "probability": 0.90185546875
                },
                {
                    "start": 1117.73,
                    "end": 1118.27,
                    "word": " of",
                    "probability": 0.87255859375
                },
                {
                    "start": 1118.27,
                    "end": 1118.43,
                    "word": " the",
                    "probability": 0.8115234375
                },
                {
                    "start": 1118.43,
                    "end": 1118.63,
                    "word": " multi",
                    "probability": 0.25146484375
                },
                {
                    "start": 1118.63,
                    "end": 1118.83,
                    "word": "-class",
                    "probability": 0.730712890625
                },
                {
                    "start": 1118.83,
                    "end": 1118.95,
                    "word": " or",
                    "probability": 0.62744140625
                },
                {
                    "start": 1118.95,
                    "end": 1119.13,
                    "word": " binary",
                    "probability": 0.822265625
                },
                {
                    "start": 1119.13,
                    "end": 1119.69,
                    "word": " class",
                    "probability": 0.91357421875
                },
                {
                    "start": 1119.69,
                    "end": 1120.21,
                    "word": ".",
                    "probability": 0.98779296875
                },
                {
                    "start": 1120.49,
                    "end": 1121.01,
                    "word": " Each",
                    "probability": 0.8603515625
                },
                {
                    "start": 1121.01,
                    "end": 1121.13,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 1121.13,
                    "end": 1121.35,
                    "word": " these",
                    "probability": 0.71435546875
                }
            ]
        },
        {
            "id": 171,
            "seek": 114130,
            "start": 1121.35,
            "end": 1128.41,
            "text": " n shadow networks will have the same networks, and it'll produce a prediction of how likely it is",
            "tokens": [
                50364,
                297,
                8576,
                9590,
                486,
                362,
                264,
                912,
                9590,
                11,
                293,
                309,
                603,
                5258,
                257,
                17630,
                295,
                577,
                3700,
                309,
                307,
                50724
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2079861130979326,
            "compression_ratio": 1.6828193832599119,
            "no_speech_prob": 0.00019407272338867188,
            "words": [
                {
                    "start": 1121.35,
                    "end": 1121.61,
                    "word": " n",
                    "probability": 0.004634857177734375
                },
                {
                    "start": 1121.61,
                    "end": 1121.85,
                    "word": " shadow",
                    "probability": 0.1982421875
                },
                {
                    "start": 1121.85,
                    "end": 1122.39,
                    "word": " networks",
                    "probability": 0.64404296875
                },
                {
                    "start": 1122.39,
                    "end": 1122.61,
                    "word": " will",
                    "probability": 0.78662109375
                },
                {
                    "start": 1122.61,
                    "end": 1122.81,
                    "word": " have",
                    "probability": 0.88134765625
                },
                {
                    "start": 1122.81,
                    "end": 1122.97,
                    "word": " the",
                    "probability": 0.8076171875
                },
                {
                    "start": 1122.97,
                    "end": 1123.15,
                    "word": " same",
                    "probability": 0.8056640625
                },
                {
                    "start": 1123.15,
                    "end": 1123.73,
                    "word": " networks",
                    "probability": 0.6181640625
                },
                {
                    "start": 1123.73,
                    "end": 1124.25,
                    "word": ",",
                    "probability": 0.16357421875
                },
                {
                    "start": 1124.25,
                    "end": 1124.61,
                    "word": " and",
                    "probability": 0.8828125
                },
                {
                    "start": 1124.61,
                    "end": 1124.91,
                    "word": " it'll",
                    "probability": 0.52734375
                },
                {
                    "start": 1124.91,
                    "end": 1125.37,
                    "word": " produce",
                    "probability": 0.7314453125
                },
                {
                    "start": 1125.55,
                    "end": 1125.87,
                    "word": " a prediction",
                    "probability": 0.8427734375
                },
                {
                    "start": 1125.87,
                    "end": 1126.51,
                    "word": " of",
                    "probability": 0.9013671875
                },
                {
                    "start": 1126.51,
                    "end": 1127.37,
                    "word": " how",
                    "probability": 0.8515625
                },
                {
                    "start": 1127.37,
                    "end": 1127.71,
                    "word": " likely",
                    "probability": 0.8212890625
                },
                {
                    "start": 1127.71,
                    "end": 1128.03,
                    "word": " it",
                    "probability": 0.9013671875
                },
                {
                    "start": 1128.03,
                    "end": 1128.41,
                    "word": " is",
                    "probability": 0.90625
                }
            ]
        },
        {
            "id": 172,
            "seek": 114130,
            "start": 1128.41,
            "end": 1134.37,
            "text": " to be in a certain class or not. If we look at the TAC network, this is where then you're feeding",
            "tokens": [
                50724,
                281,
                312,
                294,
                257,
                1629,
                1508,
                420,
                406,
                13,
                759,
                321,
                574,
                412,
                264,
                314,
                4378,
                3209,
                11,
                341,
                307,
                689,
                550,
                291,
                434,
                12919,
                51016
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2079861130979326,
            "compression_ratio": 1.6828193832599119,
            "no_speech_prob": 0.00019407272338867188,
            "words": [
                {
                    "start": 1128.41,
                    "end": 1128.81,
                    "word": " to",
                    "probability": 0.90185546875
                },
                {
                    "start": 1128.81,
                    "end": 1128.95,
                    "word": " be",
                    "probability": 0.90087890625
                },
                {
                    "start": 1128.95,
                    "end": 1129.05,
                    "word": " in",
                    "probability": 0.8759765625
                },
                {
                    "start": 1129.13,
                    "end": 1129.31,
                    "word": " a certain",
                    "probability": 0.8037109375
                },
                {
                    "start": 1129.31,
                    "end": 1129.63,
                    "word": " class",
                    "probability": 0.931640625
                },
                {
                    "start": 1129.63,
                    "end": 1129.75,
                    "word": " or",
                    "probability": 0.90380859375
                },
                {
                    "start": 1129.75,
                    "end": 1129.97,
                    "word": " not",
                    "probability": 0.892578125
                },
                {
                    "start": 1129.97,
                    "end": 1130.47,
                    "word": ".",
                    "probability": 0.98388671875
                },
                {
                    "start": 1131.47,
                    "end": 1131.97,
                    "word": " If",
                    "probability": 0.876953125
                },
                {
                    "start": 1131.97,
                    "end": 1132.07,
                    "word": " we",
                    "probability": 0.443115234375
                },
                {
                    "start": 1132.07,
                    "end": 1132.21,
                    "word": " look",
                    "probability": 0.921875
                },
                {
                    "start": 1132.21,
                    "end": 1132.33,
                    "word": " at",
                    "probability": 0.9091796875
                },
                {
                    "start": 1132.33,
                    "end": 1132.49,
                    "word": " the",
                    "probability": 0.8203125
                },
                {
                    "start": 1132.49,
                    "end": 1132.79,
                    "word": " TAC",
                    "probability": 0.31349945068359375
                },
                {
                    "start": 1132.79,
                    "end": 1133.09,
                    "word": " network",
                    "probability": 0.783203125
                },
                {
                    "start": 1133.09,
                    "end": 1133.33,
                    "word": ",",
                    "probability": 0.927734375
                },
                {
                    "start": 1133.33,
                    "end": 1133.47,
                    "word": " this",
                    "probability": 0.873046875
                },
                {
                    "start": 1133.47,
                    "end": 1133.63,
                    "word": " is",
                    "probability": 0.908203125
                },
                {
                    "start": 1133.63,
                    "end": 1133.87,
                    "word": " where",
                    "probability": 0.79150390625
                },
                {
                    "start": 1133.87,
                    "end": 1134.01,
                    "word": " then",
                    "probability": 0.634765625
                },
                {
                    "start": 1134.01,
                    "end": 1134.17,
                    "word": " you're",
                    "probability": 0.881103515625
                },
                {
                    "start": 1134.17,
                    "end": 1134.37,
                    "word": " feeding",
                    "probability": 0.75732421875
                }
            ]
        },
        {
            "id": 173,
            "seek": 114130,
            "start": 1134.37,
            "end": 1139.93,
            "text": " the predictions of the shadow networks. We're then training a very fine model to distinguish",
            "tokens": [
                51016,
                264,
                21264,
                295,
                264,
                8576,
                9590,
                13,
                492,
                434,
                550,
                3097,
                257,
                588,
                2489,
                2316,
                281,
                20206,
                51296
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2079861130979326,
            "compression_ratio": 1.6828193832599119,
            "no_speech_prob": 0.00019407272338867188,
            "words": [
                {
                    "start": 1134.37,
                    "end": 1134.57,
                    "word": " the",
                    "probability": 0.46142578125
                },
                {
                    "start": 1134.57,
                    "end": 1135.01,
                    "word": " predictions",
                    "probability": 0.732421875
                },
                {
                    "start": 1135.01,
                    "end": 1135.51,
                    "word": " of",
                    "probability": 0.904296875
                },
                {
                    "start": 1135.51,
                    "end": 1135.67,
                    "word": " the",
                    "probability": 0.81298828125
                },
                {
                    "start": 1135.67,
                    "end": 1135.87,
                    "word": " shadow",
                    "probability": 0.7607421875
                },
                {
                    "start": 1135.87,
                    "end": 1136.55,
                    "word": " networks",
                    "probability": 0.64794921875
                },
                {
                    "start": 1136.55,
                    "end": 1136.69,
                    "word": ".",
                    "probability": 0.30908203125
                },
                {
                    "start": 1136.69,
                    "end": 1137.05,
                    "word": " We're",
                    "probability": 0.864501953125
                },
                {
                    "start": 1137.05,
                    "end": 1137.29,
                    "word": " then",
                    "probability": 0.7197265625
                },
                {
                    "start": 1137.29,
                    "end": 1137.57,
                    "word": " training",
                    "probability": 0.364501953125
                },
                {
                    "start": 1137.79,
                    "end": 1138.01,
                    "word": " a very",
                    "probability": 0.6767578125
                },
                {
                    "start": 1138.01,
                    "end": 1138.49,
                    "word": " fine",
                    "probability": 0.7001953125
                },
                {
                    "start": 1138.49,
                    "end": 1139.25,
                    "word": " model",
                    "probability": 0.8037109375
                },
                {
                    "start": 1139.25,
                    "end": 1139.49,
                    "word": " to",
                    "probability": 0.90771484375
                },
                {
                    "start": 1139.49,
                    "end": 1139.93,
                    "word": " distinguish",
                    "probability": 0.767578125
                }
            ]
        },
        {
            "id": 174,
            "seek": 114130,
            "start": 1139.93,
            "end": 1145.61,
            "text": " the threshold of whether a particular observation is within a network or not. So this is very",
            "tokens": [
                51296,
                264,
                14678,
                295,
                1968,
                257,
                1729,
                14816,
                307,
                1951,
                257,
                3209,
                420,
                406,
                13,
                407,
                341,
                307,
                588,
                51588
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2079861130979326,
            "compression_ratio": 1.6828193832599119,
            "no_speech_prob": 0.00019407272338867188,
            "words": [
                {
                    "start": 1139.93,
                    "end": 1140.19,
                    "word": " the",
                    "probability": 0.724609375
                },
                {
                    "start": 1140.19,
                    "end": 1140.49,
                    "word": " threshold",
                    "probability": 0.61669921875
                },
                {
                    "start": 1140.49,
                    "end": 1140.73,
                    "word": " of",
                    "probability": 0.89599609375
                },
                {
                    "start": 1140.73,
                    "end": 1141.03,
                    "word": " whether",
                    "probability": 0.8447265625
                },
                {
                    "start": 1141.61,
                    "end": 1141.87,
                    "word": " a particular",
                    "probability": 0.80810546875
                },
                {
                    "start": 1141.87,
                    "end": 1142.33,
                    "word": " observation",
                    "probability": 0.8037109375
                },
                {
                    "start": 1142.33,
                    "end": 1142.95,
                    "word": " is",
                    "probability": 0.91064453125
                },
                {
                    "start": 1142.95,
                    "end": 1143.31,
                    "word": " within",
                    "probability": 0.75048828125
                },
                {
                    "start": 1143.45,
                    "end": 1143.65,
                    "word": " a network",
                    "probability": 0.8486328125
                },
                {
                    "start": 1143.65,
                    "end": 1143.81,
                    "word": " or",
                    "probability": 0.90478515625
                },
                {
                    "start": 1143.81,
                    "end": 1144.19,
                    "word": " not",
                    "probability": 0.8916015625
                },
                {
                    "start": 1144.19,
                    "end": 1144.69,
                    "word": ".",
                    "probability": 0.9853515625
                },
                {
                    "start": 1144.75,
                    "end": 1145.11,
                    "word": " So",
                    "probability": 0.81884765625
                },
                {
                    "start": 1145.11,
                    "end": 1145.25,
                    "word": " this",
                    "probability": 0.7734375
                },
                {
                    "start": 1145.25,
                    "end": 1145.37,
                    "word": " is",
                    "probability": 0.908203125
                },
                {
                    "start": 1145.37,
                    "end": 1145.61,
                    "word": " very",
                    "probability": 0.72705078125
                }
            ]
        },
        {
            "id": 175,
            "seek": 116720,
            "start": 1145.61,
            "end": 1151.91,
            "text": " simple. You're training samples from the shadow network. You're going to identify if it came from",
            "tokens": [
                50376,
                2199,
                13,
                509,
                434,
                3097,
                10938,
                490,
                264,
                8576,
                3209,
                13,
                509,
                434,
                516,
                281,
                5876,
                498,
                309,
                1361,
                490,
                50672
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2343750042073867,
            "compression_ratio": 1.5850622406639003,
            "no_speech_prob": 0.0022869110107421875,
            "words": [
                {
                    "start": 1145.61,
                    "end": 1145.89,
                    "word": " simple",
                    "probability": 0.00907135009765625
                },
                {
                    "start": 1145.89,
                    "end": 1145.89,
                    "word": ".",
                    "probability": 0.210205078125
                },
                {
                    "start": 1146.43,
                    "end": 1147.03,
                    "word": " You're",
                    "probability": 0.5906982421875
                },
                {
                    "start": 1147.03,
                    "end": 1147.33,
                    "word": " training",
                    "probability": 0.267822265625
                },
                {
                    "start": 1147.33,
                    "end": 1147.73,
                    "word": " samples",
                    "probability": 0.697265625
                },
                {
                    "start": 1147.73,
                    "end": 1147.99,
                    "word": " from",
                    "probability": 0.80810546875
                },
                {
                    "start": 1147.99,
                    "end": 1148.13,
                    "word": " the",
                    "probability": 0.640625
                },
                {
                    "start": 1148.13,
                    "end": 1148.35,
                    "word": " shadow",
                    "probability": 0.087890625
                },
                {
                    "start": 1148.35,
                    "end": 1148.91,
                    "word": " network",
                    "probability": 0.80859375
                },
                {
                    "start": 1148.91,
                    "end": 1149.23,
                    "word": ".",
                    "probability": 0.578125
                },
                {
                    "start": 1149.23,
                    "end": 1149.45,
                    "word": " You're",
                    "probability": 0.91259765625
                },
                {
                    "start": 1149.45,
                    "end": 1149.67,
                    "word": " going",
                    "probability": 0.85546875
                },
                {
                    "start": 1149.67,
                    "end": 1150.27,
                    "word": " to",
                    "probability": 0.90625
                },
                {
                    "start": 1150.27,
                    "end": 1150.97,
                    "word": " identify",
                    "probability": 0.8759765625
                },
                {
                    "start": 1150.97,
                    "end": 1151.35,
                    "word": " if",
                    "probability": 0.869140625
                },
                {
                    "start": 1151.35,
                    "end": 1151.47,
                    "word": " it",
                    "probability": 0.90478515625
                },
                {
                    "start": 1151.47,
                    "end": 1151.71,
                    "word": " came",
                    "probability": 0.8046875
                },
                {
                    "start": 1151.71,
                    "end": 1151.91,
                    "word": " from",
                    "probability": 0.82470703125
                }
            ]
        },
        {
            "id": 176,
            "seek": 116720,
            "start": 1151.91,
            "end": 1158.81,
            "text": " that particular shadow network or not, and that would be pretty much your label. If we go to the",
            "tokens": [
                50672,
                300,
                1729,
                8576,
                3209,
                420,
                406,
                11,
                293,
                300,
                576,
                312,
                1238,
                709,
                428,
                7645,
                13,
                759,
                321,
                352,
                281,
                264,
                51020
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2343750042073867,
            "compression_ratio": 1.5850622406639003,
            "no_speech_prob": 0.0022869110107421875,
            "words": [
                {
                    "start": 1151.91,
                    "end": 1152.07,
                    "word": " that",
                    "probability": 0.76611328125
                },
                {
                    "start": 1152.07,
                    "end": 1152.29,
                    "word": " particular",
                    "probability": 0.72265625
                },
                {
                    "start": 1152.29,
                    "end": 1152.61,
                    "word": " shadow",
                    "probability": 0.3359375
                },
                {
                    "start": 1152.61,
                    "end": 1152.99,
                    "word": " network",
                    "probability": 0.814453125
                },
                {
                    "start": 1152.99,
                    "end": 1153.25,
                    "word": " or",
                    "probability": 0.87109375
                },
                {
                    "start": 1153.25,
                    "end": 1153.57,
                    "word": " not",
                    "probability": 0.880859375
                },
                {
                    "start": 1153.57,
                    "end": 1153.87,
                    "word": ",",
                    "probability": 0.419921875
                },
                {
                    "start": 1153.87,
                    "end": 1153.97,
                    "word": " and",
                    "probability": 0.849609375
                },
                {
                    "start": 1153.97,
                    "end": 1154.09,
                    "word": " that",
                    "probability": 0.87646484375
                },
                {
                    "start": 1154.09,
                    "end": 1154.21,
                    "word": " would",
                    "probability": 0.423095703125
                },
                {
                    "start": 1154.21,
                    "end": 1154.43,
                    "word": " be",
                    "probability": 0.892578125
                },
                {
                    "start": 1154.43,
                    "end": 1154.71,
                    "word": " pretty",
                    "probability": 0.869140625
                },
                {
                    "start": 1154.71,
                    "end": 1154.91,
                    "word": " much",
                    "probability": 0.83349609375
                },
                {
                    "start": 1154.91,
                    "end": 1155.07,
                    "word": " your",
                    "probability": 0.763671875
                },
                {
                    "start": 1155.07,
                    "end": 1155.39,
                    "word": " label",
                    "probability": 0.78271484375
                },
                {
                    "start": 1155.39,
                    "end": 1155.99,
                    "word": ".",
                    "probability": 0.98828125
                },
                {
                    "start": 1157.37,
                    "end": 1157.97,
                    "word": " If",
                    "probability": 0.82861328125
                },
                {
                    "start": 1157.97,
                    "end": 1158.09,
                    "word": " we",
                    "probability": 0.9033203125
                },
                {
                    "start": 1158.09,
                    "end": 1158.23,
                    "word": " go",
                    "probability": 0.896484375
                },
                {
                    "start": 1158.23,
                    "end": 1158.55,
                    "word": " to",
                    "probability": 0.88720703125
                },
                {
                    "start": 1158.55,
                    "end": 1158.81,
                    "word": " the",
                    "probability": 0.79541015625
                }
            ]
        },
        {
            "id": 177,
            "seek": 116720,
            "start": 1158.81,
            "end": 1166.37,
            "text": " target DP, this will reveal how easy it is to implement a differential private method for your",
            "tokens": [
                51020,
                3779,
                42796,
                11,
                341,
                486,
                10658,
                577,
                1858,
                309,
                307,
                281,
                4445,
                257,
                15756,
                4551,
                3170,
                337,
                428,
                51396
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2343750042073867,
            "compression_ratio": 1.5850622406639003,
            "no_speech_prob": 0.0022869110107421875,
            "words": [
                {
                    "start": 1158.81,
                    "end": 1159.05,
                    "word": " target",
                    "probability": 0.7294921875
                },
                {
                    "start": 1159.05,
                    "end": 1159.49,
                    "word": " DP",
                    "probability": 0.50048828125
                },
                {
                    "start": 1159.49,
                    "end": 1160.11,
                    "word": ",",
                    "probability": 0.9072265625
                },
                {
                    "start": 1160.11,
                    "end": 1160.55,
                    "word": " this",
                    "probability": 0.87158203125
                },
                {
                    "start": 1160.55,
                    "end": 1161.15,
                    "word": " will",
                    "probability": 0.798828125
                },
                {
                    "start": 1161.15,
                    "end": 1162.67,
                    "word": " reveal",
                    "probability": 0.74462890625
                },
                {
                    "start": 1162.67,
                    "end": 1162.89,
                    "word": " how",
                    "probability": 0.8984375
                },
                {
                    "start": 1162.89,
                    "end": 1163.21,
                    "word": " easy",
                    "probability": 0.80419921875
                },
                {
                    "start": 1163.21,
                    "end": 1163.41,
                    "word": " it",
                    "probability": 0.90380859375
                },
                {
                    "start": 1163.41,
                    "end": 1163.63,
                    "word": " is",
                    "probability": 0.90673828125
                },
                {
                    "start": 1163.63,
                    "end": 1163.87,
                    "word": " to",
                    "probability": 0.90380859375
                },
                {
                    "start": 1163.87,
                    "end": 1164.29,
                    "word": " implement",
                    "probability": 0.763671875
                },
                {
                    "start": 1164.49,
                    "end": 1164.73,
                    "word": " a differential",
                    "probability": 0.191162109375
                },
                {
                    "start": 1164.73,
                    "end": 1165.39,
                    "word": " private",
                    "probability": 0.75390625
                },
                {
                    "start": 1165.39,
                    "end": 1165.85,
                    "word": " method",
                    "probability": 0.892578125
                },
                {
                    "start": 1165.85,
                    "end": 1166.19,
                    "word": " for",
                    "probability": 0.8935546875
                },
                {
                    "start": 1166.19,
                    "end": 1166.37,
                    "word": " your",
                    "probability": 0.509765625
                }
            ]
        },
        {
            "id": 178,
            "seek": 116720,
            "start": 1166.37,
            "end": 1171.51,
            "text": " defense mechanism if you already have a deep learning TensorFlow model in production. So all",
            "tokens": [
                51396,
                7654,
                7513,
                498,
                291,
                1217,
                362,
                257,
                2452,
                2539,
                37624,
                2316,
                294,
                4265,
                13,
                407,
                439,
                51656
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2343750042073867,
            "compression_ratio": 1.5850622406639003,
            "no_speech_prob": 0.0022869110107421875,
            "words": [
                {
                    "start": 1166.37,
                    "end": 1166.67,
                    "word": " defense",
                    "probability": 0.78125
                },
                {
                    "start": 1166.67,
                    "end": 1167.11,
                    "word": " mechanism",
                    "probability": 0.900390625
                },
                {
                    "start": 1167.11,
                    "end": 1167.47,
                    "word": " if",
                    "probability": 0.64892578125
                },
                {
                    "start": 1167.47,
                    "end": 1167.63,
                    "word": " you",
                    "probability": 0.89697265625
                },
                {
                    "start": 1167.63,
                    "end": 1167.77,
                    "word": " already",
                    "probability": 0.53125
                },
                {
                    "start": 1167.77,
                    "end": 1168.13,
                    "word": " have",
                    "probability": 0.87841796875
                },
                {
                    "start": 1168.55,
                    "end": 1168.85,
                    "word": " a deep",
                    "probability": 0.78515625
                },
                {
                    "start": 1168.85,
                    "end": 1169.09,
                    "word": " learning",
                    "probability": 0.7607421875
                },
                {
                    "start": 1169.09,
                    "end": 1169.39,
                    "word": " TensorFlow",
                    "probability": 0.71630859375
                },
                {
                    "start": 1169.39,
                    "end": 1169.77,
                    "word": " model",
                    "probability": 0.83056640625
                },
                {
                    "start": 1169.77,
                    "end": 1170.07,
                    "word": " in",
                    "probability": 0.89990234375
                },
                {
                    "start": 1170.07,
                    "end": 1170.45,
                    "word": " production",
                    "probability": 0.83349609375
                },
                {
                    "start": 1170.45,
                    "end": 1171.05,
                    "word": ".",
                    "probability": 0.9814453125
                },
                {
                    "start": 1171.17,
                    "end": 1171.29,
                    "word": " So",
                    "probability": 0.673828125
                },
                {
                    "start": 1171.29,
                    "end": 1171.51,
                    "word": " all",
                    "probability": 0.74755859375
                }
            ]
        },
        {
            "id": 179,
            "seek": 119210,
            "start": 1171.51,
            "end": 1176.71,
            "text": " you really have to do is import the TensorFlow privacy library and then update your optimizer.",
            "tokens": [
                50364,
                291,
                534,
                362,
                281,
                360,
                307,
                974,
                264,
                37624,
                11427,
                6405,
                293,
                550,
                5623,
                428,
                5028,
                6545,
                13,
                50612
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2799369762925541,
            "compression_ratio": 1.684981684981685,
            "no_speech_prob": 0.001956939697265625,
            "words": [
                {
                    "start": 1171.51,
                    "end": 1171.57,
                    "word": " you",
                    "probability": 0.067626953125
                },
                {
                    "start": 1171.57,
                    "end": 1171.83,
                    "word": " really",
                    "probability": 0.5263671875
                },
                {
                    "start": 1171.83,
                    "end": 1171.99,
                    "word": " have",
                    "probability": 0.76611328125
                },
                {
                    "start": 1171.99,
                    "end": 1172.13,
                    "word": " to",
                    "probability": 0.90673828125
                },
                {
                    "start": 1172.13,
                    "end": 1172.37,
                    "word": " do",
                    "probability": 0.91845703125
                },
                {
                    "start": 1172.37,
                    "end": 1172.63,
                    "word": " is",
                    "probability": 0.89599609375
                },
                {
                    "start": 1172.63,
                    "end": 1173.03,
                    "word": " import",
                    "probability": 0.5830078125
                },
                {
                    "start": 1173.03,
                    "end": 1173.29,
                    "word": " the",
                    "probability": 0.78076171875
                },
                {
                    "start": 1173.29,
                    "end": 1173.59,
                    "word": " TensorFlow",
                    "probability": 0.4404296875
                },
                {
                    "start": 1173.59,
                    "end": 1174.01,
                    "word": " privacy",
                    "probability": 0.55419921875
                },
                {
                    "start": 1174.01,
                    "end": 1174.45,
                    "word": " library",
                    "probability": 0.81884765625
                },
                {
                    "start": 1174.45,
                    "end": 1174.83,
                    "word": " and",
                    "probability": 0.73583984375
                },
                {
                    "start": 1174.83,
                    "end": 1175.03,
                    "word": " then",
                    "probability": 0.728515625
                },
                {
                    "start": 1175.03,
                    "end": 1175.53,
                    "word": " update",
                    "probability": 0.7548828125
                },
                {
                    "start": 1175.53,
                    "end": 1175.75,
                    "word": " your",
                    "probability": 0.7822265625
                },
                {
                    "start": 1175.75,
                    "end": 1176.33,
                    "word": " optimizer",
                    "probability": 0.877685546875
                },
                {
                    "start": 1176.33,
                    "end": 1176.71,
                    "word": ".",
                    "probability": 0.92822265625
                }
            ]
        },
        {
            "id": 180,
            "seek": 119210,
            "start": 1176.71,
            "end": 1181.61,
            "text": " So for in cell 11, you can see that the model is the same, but in cell 12, you can see that",
            "tokens": [
                50612,
                407,
                337,
                294,
                2815,
                2975,
                11,
                291,
                393,
                536,
                300,
                264,
                2316,
                307,
                264,
                912,
                11,
                457,
                294,
                2815,
                2272,
                11,
                291,
                393,
                536,
                300,
                50872
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2799369762925541,
            "compression_ratio": 1.684981684981685,
            "no_speech_prob": 0.001956939697265625,
            "words": [
                {
                    "start": 1176.71,
                    "end": 1176.71,
                    "word": " So",
                    "probability": 0.80126953125
                },
                {
                    "start": 1176.71,
                    "end": 1176.95,
                    "word": " for",
                    "probability": 0.55419921875
                },
                {
                    "start": 1176.95,
                    "end": 1177.39,
                    "word": " in",
                    "probability": 0.2176513671875
                },
                {
                    "start": 1177.39,
                    "end": 1177.51,
                    "word": " cell",
                    "probability": 0.634765625
                },
                {
                    "start": 1177.51,
                    "end": 1177.73,
                    "word": " 11",
                    "probability": 0.8896484375
                },
                {
                    "start": 1177.73,
                    "end": 1177.73,
                    "word": ",",
                    "probability": 0.7119140625
                },
                {
                    "start": 1177.73,
                    "end": 1177.85,
                    "word": " you",
                    "probability": 0.82080078125
                },
                {
                    "start": 1177.85,
                    "end": 1177.97,
                    "word": " can",
                    "probability": 0.68701171875
                },
                {
                    "start": 1177.97,
                    "end": 1178.19,
                    "word": " see",
                    "probability": 0.8173828125
                },
                {
                    "start": 1178.19,
                    "end": 1178.49,
                    "word": " that",
                    "probability": 0.87548828125
                },
                {
                    "start": 1178.49,
                    "end": 1178.73,
                    "word": " the",
                    "probability": 0.783203125
                },
                {
                    "start": 1178.73,
                    "end": 1179.07,
                    "word": " model",
                    "probability": 0.779296875
                },
                {
                    "start": 1179.07,
                    "end": 1179.43,
                    "word": " is",
                    "probability": 0.90087890625
                },
                {
                    "start": 1179.43,
                    "end": 1179.59,
                    "word": " the",
                    "probability": 0.8134765625
                },
                {
                    "start": 1179.59,
                    "end": 1179.83,
                    "word": " same",
                    "probability": 0.8134765625
                },
                {
                    "start": 1179.83,
                    "end": 1180.05,
                    "word": ",",
                    "probability": 0.587890625
                },
                {
                    "start": 1180.05,
                    "end": 1180.17,
                    "word": " but",
                    "probability": 0.88330078125
                },
                {
                    "start": 1180.17,
                    "end": 1180.41,
                    "word": " in",
                    "probability": 0.80517578125
                },
                {
                    "start": 1180.41,
                    "end": 1180.63,
                    "word": " cell",
                    "probability": 0.82421875
                },
                {
                    "start": 1180.63,
                    "end": 1181.01,
                    "word": " 12",
                    "probability": 0.904296875
                },
                {
                    "start": 1181.01,
                    "end": 1181.09,
                    "word": ",",
                    "probability": 0.658203125
                },
                {
                    "start": 1181.09,
                    "end": 1181.15,
                    "word": " you",
                    "probability": 0.83837890625
                },
                {
                    "start": 1181.15,
                    "end": 1181.21,
                    "word": " can",
                    "probability": 0.7158203125
                },
                {
                    "start": 1181.21,
                    "end": 1181.35,
                    "word": " see",
                    "probability": 0.81982421875
                },
                {
                    "start": 1181.35,
                    "end": 1181.61,
                    "word": " that",
                    "probability": 0.87255859375
                }
            ]
        },
        {
            "id": 181,
            "seek": 119210,
            "start": 1181.61,
            "end": 1186.45,
            "text": " the only thing that changes is the optimizer, which is a DP gradient descent constant optimizer",
            "tokens": [
                50872,
                264,
                787,
                551,
                300,
                2962,
                307,
                264,
                5028,
                6545,
                11,
                597,
                307,
                257,
                42796,
                16235,
                23475,
                5754,
                5028,
                6545,
                51136
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2799369762925541,
            "compression_ratio": 1.684981684981685,
            "no_speech_prob": 0.001956939697265625,
            "words": [
                {
                    "start": 1181.61,
                    "end": 1181.99,
                    "word": " the",
                    "probability": 0.76171875
                },
                {
                    "start": 1181.99,
                    "end": 1182.07,
                    "word": " only",
                    "probability": 0.83203125
                },
                {
                    "start": 1182.07,
                    "end": 1182.23,
                    "word": " thing",
                    "probability": 0.79638671875
                },
                {
                    "start": 1182.23,
                    "end": 1182.37,
                    "word": " that",
                    "probability": 0.8837890625
                },
                {
                    "start": 1182.37,
                    "end": 1182.59,
                    "word": " changes",
                    "probability": 0.7548828125
                },
                {
                    "start": 1182.59,
                    "end": 1182.83,
                    "word": " is",
                    "probability": 0.88427734375
                },
                {
                    "start": 1182.83,
                    "end": 1182.97,
                    "word": " the",
                    "probability": 0.80908203125
                },
                {
                    "start": 1182.97,
                    "end": 1183.81,
                    "word": " optimizer",
                    "probability": 0.913330078125
                },
                {
                    "start": 1183.81,
                    "end": 1183.95,
                    "word": ",",
                    "probability": 0.76904296875
                },
                {
                    "start": 1183.95,
                    "end": 1184.05,
                    "word": " which",
                    "probability": 0.90380859375
                },
                {
                    "start": 1184.05,
                    "end": 1184.13,
                    "word": " is",
                    "probability": 0.6748046875
                },
                {
                    "start": 1184.27,
                    "end": 1184.51,
                    "word": " a DP",
                    "probability": 0.409423828125
                },
                {
                    "start": 1184.51,
                    "end": 1184.95,
                    "word": " gradient",
                    "probability": 0.85400390625
                },
                {
                    "start": 1184.95,
                    "end": 1185.25,
                    "word": " descent",
                    "probability": 0.654296875
                },
                {
                    "start": 1185.25,
                    "end": 1185.83,
                    "word": " constant",
                    "probability": 0.39208984375
                },
                {
                    "start": 1185.83,
                    "end": 1186.45,
                    "word": " optimizer",
                    "probability": 0.921630859375
                }
            ]
        },
        {
            "id": 182,
            "seek": 119210,
            "start": 1186.45,
            "end": 1191.81,
            "text": " with four inputs that you would have to fine-tune. Dial-to-norm clip, noise multiplier,",
            "tokens": [
                51136,
                365,
                1451,
                15743,
                300,
                291,
                576,
                362,
                281,
                2489,
                12,
                83,
                2613,
                13,
                29658,
                12,
                1353,
                12,
                13403,
                7353,
                11,
                5658,
                44106,
                11,
                51368
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2799369762925541,
            "compression_ratio": 1.684981684981685,
            "no_speech_prob": 0.001956939697265625,
            "words": [
                {
                    "start": 1186.45,
                    "end": 1187.21,
                    "word": " with",
                    "probability": 0.62109375
                },
                {
                    "start": 1187.21,
                    "end": 1187.67,
                    "word": " four",
                    "probability": 0.85595703125
                },
                {
                    "start": 1187.67,
                    "end": 1188.07,
                    "word": " inputs",
                    "probability": 0.76611328125
                },
                {
                    "start": 1188.07,
                    "end": 1188.23,
                    "word": " that",
                    "probability": 0.87744140625
                },
                {
                    "start": 1188.23,
                    "end": 1188.33,
                    "word": " you",
                    "probability": 0.89306640625
                },
                {
                    "start": 1188.33,
                    "end": 1188.47,
                    "word": " would",
                    "probability": 0.8466796875
                },
                {
                    "start": 1188.47,
                    "end": 1188.69,
                    "word": " have",
                    "probability": 0.8798828125
                },
                {
                    "start": 1188.69,
                    "end": 1188.85,
                    "word": " to",
                    "probability": 0.884765625
                },
                {
                    "start": 1188.85,
                    "end": 1189.03,
                    "word": " fine",
                    "probability": 0.8701171875
                },
                {
                    "start": 1189.03,
                    "end": 1189.71,
                    "word": "-tune",
                    "probability": 0.6904703776041666
                },
                {
                    "start": 1189.71,
                    "end": 1189.85,
                    "word": ".",
                    "probability": 0.52685546875
                },
                {
                    "start": 1189.85,
                    "end": 1190.03,
                    "word": " Dial",
                    "probability": 0.1654052734375
                },
                {
                    "start": 1190.03,
                    "end": 1190.19,
                    "word": "-to",
                    "probability": 0.6612548828125
                },
                {
                    "start": 1190.19,
                    "end": 1190.35,
                    "word": "-norm",
                    "probability": 0.905517578125
                },
                {
                    "start": 1190.35,
                    "end": 1190.65,
                    "word": " clip",
                    "probability": 0.57958984375
                },
                {
                    "start": 1190.65,
                    "end": 1190.79,
                    "word": ",",
                    "probability": 0.97998046875
                },
                {
                    "start": 1190.79,
                    "end": 1190.97,
                    "word": " noise",
                    "probability": 0.88525390625
                },
                {
                    "start": 1190.97,
                    "end": 1191.41,
                    "word": " multiplier",
                    "probability": 0.59619140625
                },
                {
                    "start": 1191.41,
                    "end": 1191.81,
                    "word": ",",
                    "probability": 0.98828125
                }
            ]
        },
        {
            "id": 183,
            "seek": 119210,
            "start": 1191.81,
            "end": 1196.41,
            "text": " non-micro-batches, and learning rate. And then you have your differential private method.",
            "tokens": [
                51368,
                2107,
                12,
                13195,
                340,
                12,
                65,
                852,
                279,
                11,
                293,
                2539,
                3314,
                13,
                400,
                550,
                291,
                362,
                428,
                15756,
                4551,
                3170,
                13,
                51588
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2799369762925541,
            "compression_ratio": 1.684981684981685,
            "no_speech_prob": 0.001956939697265625,
            "words": [
                {
                    "start": 1191.81,
                    "end": 1191.93,
                    "word": " non",
                    "probability": 0.6044921875
                },
                {
                    "start": 1191.93,
                    "end": 1192.23,
                    "word": "-micro",
                    "probability": 0.85791015625
                },
                {
                    "start": 1192.23,
                    "end": 1192.59,
                    "word": "-batches",
                    "probability": 0.7314453125
                },
                {
                    "start": 1192.59,
                    "end": 1192.67,
                    "word": ",",
                    "probability": 0.796875
                },
                {
                    "start": 1192.67,
                    "end": 1192.77,
                    "word": " and",
                    "probability": 0.89453125
                },
                {
                    "start": 1192.77,
                    "end": 1192.97,
                    "word": " learning",
                    "probability": 0.8134765625
                },
                {
                    "start": 1192.97,
                    "end": 1193.19,
                    "word": " rate",
                    "probability": 0.865234375
                },
                {
                    "start": 1193.19,
                    "end": 1193.53,
                    "word": ".",
                    "probability": 0.908203125
                },
                {
                    "start": 1193.53,
                    "end": 1193.79,
                    "word": " And",
                    "probability": 0.8486328125
                },
                {
                    "start": 1193.79,
                    "end": 1194.17,
                    "word": " then",
                    "probability": 0.72705078125
                },
                {
                    "start": 1194.17,
                    "end": 1194.33,
                    "word": " you",
                    "probability": 0.82421875
                },
                {
                    "start": 1194.33,
                    "end": 1194.55,
                    "word": " have",
                    "probability": 0.87890625
                },
                {
                    "start": 1194.55,
                    "end": 1194.83,
                    "word": " your",
                    "probability": 0.80029296875
                },
                {
                    "start": 1194.83,
                    "end": 1195.23,
                    "word": " differential",
                    "probability": 0.66357421875
                },
                {
                    "start": 1195.23,
                    "end": 1195.59,
                    "word": " private",
                    "probability": 0.7353515625
                },
                {
                    "start": 1195.59,
                    "end": 1195.93,
                    "word": " method",
                    "probability": 0.87744140625
                },
                {
                    "start": 1195.93,
                    "end": 1196.41,
                    "word": ".",
                    "probability": 0.9931640625
                }
            ]
        },
        {
            "id": 184,
            "seek": 121992,
            "start": 1199.17,
            "end": 1206.63,
            "text": " And then if we go back to the first one, here we show results of the different number of shadow",
            "tokens": [
                50408,
                400,
                550,
                498,
                321,
                352,
                646,
                281,
                264,
                700,
                472,
                11,
                510,
                321,
                855,
                3542,
                295,
                264,
                819,
                1230,
                295,
                8576,
                50812
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34907279106286854,
            "compression_ratio": 1.794392523364486,
            "no_speech_prob": 0.0011339187622070312,
            "words": [
                {
                    "start": 1199.17,
                    "end": 1199.73,
                    "word": " And",
                    "probability": 0.52197265625
                },
                {
                    "start": 1199.73,
                    "end": 1200.03,
                    "word": " then",
                    "probability": 0.7294921875
                },
                {
                    "start": 1200.03,
                    "end": 1200.95,
                    "word": " if",
                    "probability": 0.68896484375
                },
                {
                    "start": 1200.95,
                    "end": 1201.05,
                    "word": " we",
                    "probability": 0.73779296875
                },
                {
                    "start": 1201.05,
                    "end": 1201.17,
                    "word": " go",
                    "probability": 0.888671875
                },
                {
                    "start": 1201.17,
                    "end": 1201.39,
                    "word": " back",
                    "probability": 0.81982421875
                },
                {
                    "start": 1201.39,
                    "end": 1201.67,
                    "word": " to",
                    "probability": 0.90576171875
                },
                {
                    "start": 1201.67,
                    "end": 1201.99,
                    "word": " the",
                    "probability": 0.818359375
                },
                {
                    "start": 1201.99,
                    "end": 1202.23,
                    "word": " first",
                    "probability": 0.80322265625
                },
                {
                    "start": 1202.23,
                    "end": 1202.89,
                    "word": " one",
                    "probability": 0.8115234375
                },
                {
                    "start": 1202.89,
                    "end": 1203.23,
                    "word": ",",
                    "probability": 0.56103515625
                },
                {
                    "start": 1203.23,
                    "end": 1203.63,
                    "word": " here",
                    "probability": 0.7900390625
                },
                {
                    "start": 1203.63,
                    "end": 1203.91,
                    "word": " we",
                    "probability": 0.70654296875
                },
                {
                    "start": 1203.91,
                    "end": 1204.07,
                    "word": " show",
                    "probability": 0.78466796875
                },
                {
                    "start": 1204.07,
                    "end": 1204.53,
                    "word": " results",
                    "probability": 0.65087890625
                },
                {
                    "start": 1204.53,
                    "end": 1204.99,
                    "word": " of",
                    "probability": 0.8916015625
                },
                {
                    "start": 1204.99,
                    "end": 1205.17,
                    "word": " the",
                    "probability": 0.74755859375
                },
                {
                    "start": 1205.17,
                    "end": 1205.75,
                    "word": " different",
                    "probability": 0.70703125
                },
                {
                    "start": 1205.75,
                    "end": 1206.05,
                    "word": " number",
                    "probability": 0.822265625
                },
                {
                    "start": 1206.05,
                    "end": 1206.29,
                    "word": " of",
                    "probability": 0.8984375
                },
                {
                    "start": 1206.29,
                    "end": 1206.63,
                    "word": " shadow",
                    "probability": 0.78857421875
                }
            ]
        },
        {
            "id": 185,
            "seek": 121992,
            "start": 1206.63,
            "end": 1211.55,
            "text": " networks with the different number of classes, 10 classes, one-inch classes, and the binary two",
            "tokens": [
                50812,
                9590,
                365,
                264,
                819,
                1230,
                295,
                5359,
                11,
                1266,
                5359,
                11,
                472,
                12,
                12415,
                5359,
                11,
                293,
                264,
                17434,
                732,
                51052
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34907279106286854,
            "compression_ratio": 1.794392523364486,
            "no_speech_prob": 0.0011339187622070312,
            "words": [
                {
                    "start": 1206.63,
                    "end": 1207.23,
                    "word": " networks",
                    "probability": 0.356689453125
                },
                {
                    "start": 1207.23,
                    "end": 1207.81,
                    "word": " with",
                    "probability": 0.60205078125
                },
                {
                    "start": 1207.81,
                    "end": 1207.95,
                    "word": " the",
                    "probability": 0.395263671875
                },
                {
                    "start": 1207.95,
                    "end": 1208.15,
                    "word": " different",
                    "probability": 0.70458984375
                },
                {
                    "start": 1208.15,
                    "end": 1208.49,
                    "word": " number",
                    "probability": 0.8671875
                },
                {
                    "start": 1208.49,
                    "end": 1208.99,
                    "word": " of",
                    "probability": 0.90234375
                },
                {
                    "start": 1208.99,
                    "end": 1209.35,
                    "word": " classes",
                    "probability": 0.7509765625
                },
                {
                    "start": 1209.35,
                    "end": 1209.59,
                    "word": ",",
                    "probability": 0.615234375
                },
                {
                    "start": 1209.59,
                    "end": 1209.75,
                    "word": " 10",
                    "probability": 0.81005859375
                },
                {
                    "start": 1209.75,
                    "end": 1210.07,
                    "word": " classes",
                    "probability": 0.748046875
                },
                {
                    "start": 1210.07,
                    "end": 1210.17,
                    "word": ",",
                    "probability": 0.9072265625
                },
                {
                    "start": 1210.17,
                    "end": 1210.29,
                    "word": " one",
                    "probability": 0.537109375
                },
                {
                    "start": 1210.29,
                    "end": 1210.43,
                    "word": "-inch",
                    "probability": 0.4881591796875
                },
                {
                    "start": 1210.43,
                    "end": 1210.71,
                    "word": " classes",
                    "probability": 0.8173828125
                },
                {
                    "start": 1210.71,
                    "end": 1210.87,
                    "word": ",",
                    "probability": 0.509765625
                },
                {
                    "start": 1210.87,
                    "end": 1210.91,
                    "word": " and",
                    "probability": 0.888671875
                },
                {
                    "start": 1210.91,
                    "end": 1211.03,
                    "word": " the",
                    "probability": 0.7412109375
                },
                {
                    "start": 1211.03,
                    "end": 1211.29,
                    "word": " binary",
                    "probability": 0.82861328125
                },
                {
                    "start": 1211.29,
                    "end": 1211.55,
                    "word": " two",
                    "probability": 0.619140625
                }
            ]
        },
        {
            "id": 186,
            "seek": 121992,
            "start": 1211.55,
            "end": 1215.63,
            "text": " classes, and whether it was differential private or not. And you can see the different types of",
            "tokens": [
                51052,
                5359,
                11,
                293,
                1968,
                309,
                390,
                15756,
                4551,
                420,
                406,
                13,
                400,
                291,
                393,
                536,
                264,
                819,
                3467,
                295,
                51260
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34907279106286854,
            "compression_ratio": 1.794392523364486,
            "no_speech_prob": 0.0011339187622070312,
            "words": [
                {
                    "start": 1211.55,
                    "end": 1211.89,
                    "word": " classes",
                    "probability": 0.759765625
                },
                {
                    "start": 1211.89,
                    "end": 1212.01,
                    "word": ",",
                    "probability": 0.57958984375
                },
                {
                    "start": 1212.01,
                    "end": 1212.17,
                    "word": " and",
                    "probability": 0.890625
                },
                {
                    "start": 1212.17,
                    "end": 1212.37,
                    "word": " whether",
                    "probability": 0.8349609375
                },
                {
                    "start": 1212.37,
                    "end": 1212.53,
                    "word": " it",
                    "probability": 0.833984375
                },
                {
                    "start": 1212.53,
                    "end": 1212.65,
                    "word": " was",
                    "probability": 0.89892578125
                },
                {
                    "start": 1212.65,
                    "end": 1212.97,
                    "word": " differential",
                    "probability": 0.1800537109375
                },
                {
                    "start": 1212.97,
                    "end": 1213.29,
                    "word": " private",
                    "probability": 0.0784912109375
                },
                {
                    "start": 1213.29,
                    "end": 1213.45,
                    "word": " or",
                    "probability": 0.890625
                },
                {
                    "start": 1213.45,
                    "end": 1213.71,
                    "word": " not",
                    "probability": 0.896484375
                },
                {
                    "start": 1213.71,
                    "end": 1214.17,
                    "word": ".",
                    "probability": 0.86474609375
                },
                {
                    "start": 1214.17,
                    "end": 1214.25,
                    "word": " And",
                    "probability": 0.8037109375
                },
                {
                    "start": 1214.25,
                    "end": 1214.37,
                    "word": " you",
                    "probability": 0.403564453125
                },
                {
                    "start": 1214.37,
                    "end": 1214.49,
                    "word": " can",
                    "probability": 0.59765625
                },
                {
                    "start": 1214.49,
                    "end": 1214.71,
                    "word": " see",
                    "probability": 0.810546875
                },
                {
                    "start": 1214.71,
                    "end": 1214.89,
                    "word": " the",
                    "probability": 0.779296875
                },
                {
                    "start": 1214.89,
                    "end": 1215.19,
                    "word": " different",
                    "probability": 0.697265625
                },
                {
                    "start": 1215.19,
                    "end": 1215.47,
                    "word": " types",
                    "probability": 0.74755859375
                },
                {
                    "start": 1215.47,
                    "end": 1215.63,
                    "word": " of",
                    "probability": 0.90673828125
                }
            ]
        },
        {
            "id": 187,
            "seek": 121992,
            "start": 1215.63,
            "end": 1224.23,
            "text": " accuracy, test accuracy, and then the recall and precision rate as well. This is just a summary.",
            "tokens": [
                51260,
                14170,
                11,
                1500,
                14170,
                11,
                293,
                550,
                264,
                9901,
                293,
                18356,
                3314,
                382,
                731,
                13,
                639,
                307,
                445,
                257,
                12691,
                13,
                51688
            ],
            "temperature": 0.0,
            "avg_logprob": -0.34907279106286854,
            "compression_ratio": 1.794392523364486,
            "no_speech_prob": 0.0011339187622070312,
            "words": [
                {
                    "start": 1215.63,
                    "end": 1216.23,
                    "word": " accuracy",
                    "probability": 0.79248046875
                },
                {
                    "start": 1216.23,
                    "end": 1216.61,
                    "word": ",",
                    "probability": 0.8916015625
                },
                {
                    "start": 1216.61,
                    "end": 1216.87,
                    "word": " test",
                    "probability": 0.68505859375
                },
                {
                    "start": 1216.87,
                    "end": 1217.27,
                    "word": " accuracy",
                    "probability": 0.6943359375
                },
                {
                    "start": 1217.27,
                    "end": 1217.61,
                    "word": ",",
                    "probability": 0.8603515625
                },
                {
                    "start": 1217.61,
                    "end": 1217.65,
                    "word": " and",
                    "probability": 0.8896484375
                },
                {
                    "start": 1217.65,
                    "end": 1218.31,
                    "word": " then",
                    "probability": 0.73681640625
                },
                {
                    "start": 1218.31,
                    "end": 1218.73,
                    "word": " the",
                    "probability": 0.79150390625
                },
                {
                    "start": 1218.73,
                    "end": 1219.57,
                    "word": " recall",
                    "probability": 0.7265625
                },
                {
                    "start": 1219.57,
                    "end": 1219.77,
                    "word": " and",
                    "probability": 0.86962890625
                },
                {
                    "start": 1219.77,
                    "end": 1220.09,
                    "word": " precision",
                    "probability": 0.7666015625
                },
                {
                    "start": 1220.09,
                    "end": 1220.39,
                    "word": " rate",
                    "probability": 0.890625
                },
                {
                    "start": 1220.39,
                    "end": 1220.67,
                    "word": " as",
                    "probability": 0.86669921875
                },
                {
                    "start": 1220.67,
                    "end": 1220.95,
                    "word": " well",
                    "probability": 0.875
                },
                {
                    "start": 1220.95,
                    "end": 1221.51,
                    "word": ".",
                    "probability": 0.99609375
                },
                {
                    "start": 1222.47,
                    "end": 1223.03,
                    "word": " This",
                    "probability": 0.5849609375
                },
                {
                    "start": 1223.03,
                    "end": 1223.21,
                    "word": " is",
                    "probability": 0.90869140625
                },
                {
                    "start": 1223.21,
                    "end": 1223.47,
                    "word": " just",
                    "probability": 0.87353515625
                },
                {
                    "start": 1223.81,
                    "end": 1223.99,
                    "word": " a summary",
                    "probability": 0.77587890625
                },
                {
                    "start": 1223.99,
                    "end": 1224.23,
                    "word": ".",
                    "probability": 0.4501953125
                }
            ]
        },
        {
            "id": 188,
            "seek": 123840,
            "start": 1224.23,
            "end": 1228.35,
            "text": " If you're interested, you can pause here and look at it for further details.",
            "tokens": [
                50364,
                759,
                291,
                434,
                3102,
                11,
                291,
                393,
                10465,
                510,
                293,
                574,
                412,
                309,
                337,
                3052,
                4365,
                13,
                50544
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2566406294703484,
            "compression_ratio": 1.4363636363636363,
            "no_speech_prob": 0.0012254714965820312,
            "words": [
                {
                    "start": 1224.23,
                    "end": 1224.35,
                    "word": " If",
                    "probability": 0.26123046875
                },
                {
                    "start": 1224.35,
                    "end": 1224.49,
                    "word": " you're",
                    "probability": 0.8388671875
                },
                {
                    "start": 1224.49,
                    "end": 1224.75,
                    "word": " interested",
                    "probability": 0.6669921875
                },
                {
                    "start": 1224.75,
                    "end": 1224.93,
                    "word": ",",
                    "probability": 0.58251953125
                },
                {
                    "start": 1224.93,
                    "end": 1225.01,
                    "word": " you",
                    "probability": 0.70166015625
                },
                {
                    "start": 1225.01,
                    "end": 1225.11,
                    "word": " can",
                    "probability": 0.62451171875
                },
                {
                    "start": 1225.11,
                    "end": 1225.35,
                    "word": " pause",
                    "probability": 0.828125
                },
                {
                    "start": 1225.35,
                    "end": 1225.59,
                    "word": " here",
                    "probability": 0.80224609375
                },
                {
                    "start": 1225.59,
                    "end": 1225.91,
                    "word": " and",
                    "probability": 0.85791015625
                },
                {
                    "start": 1225.91,
                    "end": 1226.29,
                    "word": " look",
                    "probability": 0.9052734375
                },
                {
                    "start": 1226.29,
                    "end": 1226.43,
                    "word": " at",
                    "probability": 0.87451171875
                },
                {
                    "start": 1226.43,
                    "end": 1226.59,
                    "word": " it",
                    "probability": 0.8896484375
                },
                {
                    "start": 1226.59,
                    "end": 1226.77,
                    "word": " for",
                    "probability": 0.88916015625
                },
                {
                    "start": 1226.77,
                    "end": 1227.41,
                    "word": " further",
                    "probability": 0.83349609375
                },
                {
                    "start": 1227.41,
                    "end": 1227.91,
                    "word": " details",
                    "probability": 0.71142578125
                },
                {
                    "start": 1227.91,
                    "end": 1228.35,
                    "word": ".",
                    "probability": 0.98291015625
                }
            ]
        },
        {
            "id": 189,
            "seek": 123840,
            "start": 1235.59,
            "end": 1240.19,
            "text": " Okay, so this is the end of our presentation. Thank you so much for your time.",
            "tokens": [
                50672,
                1033,
                11,
                370,
                341,
                307,
                264,
                917,
                295,
                527,
                5860,
                13,
                1044,
                291,
                370,
                709,
                337,
                428,
                565,
                13,
                50896
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2566406294703484,
            "compression_ratio": 1.4363636363636363,
            "no_speech_prob": 0.0012254714965820312,
            "words": [
                {
                    "start": 1235.59,
                    "end": 1236.03,
                    "word": " Okay",
                    "probability": 0.270263671875
                },
                {
                    "start": 1236.03,
                    "end": 1236.27,
                    "word": ",",
                    "probability": 0.64599609375
                },
                {
                    "start": 1236.27,
                    "end": 1236.49,
                    "word": " so",
                    "probability": 0.875
                },
                {
                    "start": 1236.49,
                    "end": 1236.73,
                    "word": " this",
                    "probability": 0.86865234375
                },
                {
                    "start": 1236.73,
                    "end": 1236.87,
                    "word": " is",
                    "probability": 0.9052734375
                },
                {
                    "start": 1236.87,
                    "end": 1236.99,
                    "word": " the",
                    "probability": 0.8134765625
                },
                {
                    "start": 1236.99,
                    "end": 1237.13,
                    "word": " end",
                    "probability": 0.80615234375
                },
                {
                    "start": 1237.13,
                    "end": 1237.27,
                    "word": " of",
                    "probability": 0.9130859375
                },
                {
                    "start": 1237.27,
                    "end": 1237.43,
                    "word": " our",
                    "probability": 0.818359375
                },
                {
                    "start": 1237.43,
                    "end": 1238.09,
                    "word": " presentation",
                    "probability": 0.759765625
                },
                {
                    "start": 1238.09,
                    "end": 1238.41,
                    "word": ".",
                    "probability": 0.8603515625
                },
                {
                    "start": 1238.41,
                    "end": 1238.57,
                    "word": " Thank",
                    "probability": 0.87841796875
                },
                {
                    "start": 1238.57,
                    "end": 1238.71,
                    "word": " you",
                    "probability": 0.89404296875
                },
                {
                    "start": 1238.71,
                    "end": 1238.89,
                    "word": " so",
                    "probability": 0.904296875
                },
                {
                    "start": 1238.89,
                    "end": 1239.05,
                    "word": " much",
                    "probability": 0.8232421875
                },
                {
                    "start": 1239.05,
                    "end": 1239.23,
                    "word": " for",
                    "probability": 0.89892578125
                },
                {
                    "start": 1239.23,
                    "end": 1239.43,
                    "word": " your",
                    "probability": 0.8154296875
                },
                {
                    "start": 1239.43,
                    "end": 1239.85,
                    "word": " time",
                    "probability": 0.8095703125
                },
                {
                    "start": 1239.85,
                    "end": 1240.19,
                    "word": ".",
                    "probability": 0.93212890625
                }
            ]
        },
        {
            "id": 190,
            "seek": 123840,
            "start": 1240.19,
            "end": 1245.35,
            "text": " If you have questions, please reach out to our emails included in the slide deck.",
            "tokens": [
                50896,
                759,
                291,
                362,
                1651,
                11,
                1767,
                2524,
                484,
                281,
                527,
                12524,
                5556,
                294,
                264,
                4137,
                9341,
                13,
                51156
            ],
            "temperature": 0.0,
            "avg_logprob": -0.2566406294703484,
            "compression_ratio": 1.4363636363636363,
            "no_speech_prob": 0.0012254714965820312,
            "words": [
                {
                    "start": 1240.19,
                    "end": 1240.47,
                    "word": " If",
                    "probability": 0.8994140625
                },
                {
                    "start": 1240.47,
                    "end": 1240.61,
                    "word": " you",
                    "probability": 0.89892578125
                },
                {
                    "start": 1240.61,
                    "end": 1240.77,
                    "word": " have",
                    "probability": 0.88037109375
                },
                {
                    "start": 1240.77,
                    "end": 1241.25,
                    "word": " questions",
                    "probability": 0.814453125
                },
                {
                    "start": 1241.25,
                    "end": 1241.51,
                    "word": ",",
                    "probability": 0.91357421875
                },
                {
                    "start": 1241.51,
                    "end": 1241.73,
                    "word": " please",
                    "probability": 0.7001953125
                },
                {
                    "start": 1241.73,
                    "end": 1242.47,
                    "word": " reach",
                    "probability": 0.76220703125
                },
                {
                    "start": 1242.47,
                    "end": 1242.67,
                    "word": " out",
                    "probability": 0.81494140625
                },
                {
                    "start": 1242.67,
                    "end": 1242.81,
                    "word": " to",
                    "probability": 0.90283203125
                },
                {
                    "start": 1242.81,
                    "end": 1243.03,
                    "word": " our",
                    "probability": 0.8193359375
                },
                {
                    "start": 1243.03,
                    "end": 1243.43,
                    "word": " emails",
                    "probability": 0.72314453125
                },
                {
                    "start": 1243.43,
                    "end": 1243.91,
                    "word": " included",
                    "probability": 0.7685546875
                },
                {
                    "start": 1243.91,
                    "end": 1244.25,
                    "word": " in",
                    "probability": 0.89892578125
                },
                {
                    "start": 1244.25,
                    "end": 1244.45,
                    "word": " the",
                    "probability": 0.6904296875
                },
                {
                    "start": 1244.45,
                    "end": 1244.81,
                    "word": " slide",
                    "probability": 0.806640625
                },
                {
                    "start": 1244.81,
                    "end": 1245.09,
                    "word": " deck",
                    "probability": 0.9375
                },
                {
                    "start": 1245.09,
                    "end": 1245.35,
                    "word": ".",
                    "probability": 0.99462890625
                }
            ]
        }
    ],
    "text": " Sorry about the no sound, people. I will figure this out. Every day, it just decides to grab the wrong audio input. So let's go back to the beginning and see if this works. Hi everyone. Today's talk is about machine learning privacy. The title of the talk is secrets are lies, sharing is caring, privacy is theft. My name is Nahid Farhadi and I'm a software developer in Capital One. I will be giving this talk with my colleague Vincent Pham. The outline of our talk is as below. We will give an introduction on the importance of privacy and we will talk about privacy attacks, we propose some defense techniques, and finally we will show a demo of an attack as well as defense techniques. Machine learning needs data. So what if this data has sensitive information like medical data or financial data? Even if we trust the algorithms that are generating the model or model developers, sometimes just using the statistics by looking at the output of a model, we can find out some information about the input of the model. This type of threat to ML can be black box or white box, meaning that the adversary doesn't necessarily need to have any access to the model specifics. So what is really privacy? Privacy means that if we are using some data from our users for training a model, but just by looking at the output of the model, no one should be able to get information on the input of the model. There is a trade-off between privacy and utility. If we have too much privacy, then it will be very difficult to use the model. If we don't have much privacy, then it will be useful, but then we will endanger our customers. PPML is there to define or find the optimized point between utility and privacy. How can we sacrifice less by getting enough privacy? My colleague will be talking about privacy attacks next. In terms of privacy attack, there are two main categories. One is the classical, where the private data is in the train, where it's in the raw set. And there are the ML-enabled one, which I'll be focusing more on. In the next slide, you can see that there are several attack surfaces that attackers can utilize to perform a privacy attack. The first is the physical domain, in terms of network protection, it could be attack network traffic. Then there's the digital representation, which is the TCP dump. There's also the machine learning model, which is just the model itself, where you have an input and do a prediction on the output. In terms of this, it's the attack probability. And then you have the physical domain, which is the shutdown infrastructure. For this talk, we'll focus on the machine learning model. In the next slide, we will see our first type of attack, which is the linkage attack. With this attack, in the early 90s, Tina Sweeney found that she was able to identify the identification of the governor of Massachusetts by linking his health records to a voter information just by using three variables, his gender, date of birth, and a zip code. In the next slide, she was able to identify that using these three variables alone, she was able to uniquely identify 87% of the population in the U.S. using the census, the 1990 census. In 2000, Philip Goh identified that using these three records, he was able to identify only 64% of the population. This could be a result of people urbanizing more, moving from the rural areas to the cities. But this provides an important point where using just very simple features, attackers can identify back to the original source. In the next slide, we look at another popular linkage attack. In 2006, Netflix released a public data set, a competition where they're offering a prize to improve their recommendation. And their FAQs denoted that there is no private information in the data set itself, but researchers have found that if they were able to link the Netflix accounts back to the IMDb, just by using the name of the person on the IMDb, their public dates, and public ratings, they'll identify personable information back to Netflix. You might be wondering why this might be damaging since the people tend to put their public information on IMDb. This could be because they might be biasing or selecting a set of records to disclose on IMDb, but not on Netflix recommendation. For example, they were able to identify that for a particular person, they were able to identify just based on his Netflix record, that he might have a voting preference from his voting record on Power and Terror, Norm Chomsky and Our Times, and Periodic 9-11. They might be able to infer on his political preference based on the Jesus of Nazareth and Gospel of John ratings, and also his eating habit based on his recommendations on supersizing. In the next slide, we look at a new type of attack called the reconstruction attack. This is more prone to models that tend to keep the training information in the structure itself, such as near-neighbor classifier and kernel-based SVM. But this is also prone to neural networks as well, as we can see in the next slide, where in production environments, researchers and even implementers tend to reuse neural network models just because of something called style transfer, or style transfer where they're able to utilize one model and create a new prediction from that same model using the inputs and architecture of it, and reconstruct back to the original input. In the next slide, we also see a similar type of attack called model inversion attack, where in this case, a feature such as an image is fed into a neural network model and produces a set of probabilities, and then the researcher can use these probabilities and name and reconstruct an image back to the real thing. For example, in the right side of the slide, you can see that in the left image, there's this reconstructed image, and in the right is the original image, and they're very similar to each other. In the next slide, we also have a membership inference attack, where an adversary can generate multiple shadow network models that resembles the production environment, produce the probabilities from these models, create a new model called the attack model to feed in these probabilities, and learn what observations are trained by the production model and what are not trained by the production model, identify what are the members and what are not the members. In the next slide, we also have a core attack. This is more prone to models such as LSTM, or models that learn from a time series of data points. For this example, you might have a model that learns from social security, and it does an autocomplete, and one of the autocomplete could be some sort of sensitive information, such as the phone number or the social security of a person. This is more prone to maybe like if you're over-memorizing from the neural network model itself, where the records might appear at least five times. In the next slide, we have a summary of different privacy attacks that I've talked about earlier on. You could take a pause here and just look through the difference, compare the different approaches, and look at the examples, but we can move on if you're ready to move on. Next slide. All right, so now we're really talking about methods and techniques to preserve privacy. To preserve privacy, we have four main points to protect. The training data and input and output of the model shouldn't be visible to anyone who is not authorized to visit them. And finally, the model privacy, which makes sure that the model cannot be stolen by any malicious party. Two main categories to preserve privacy are secure computation and privacy preservation. In case of secure computation, we transform or distribute our computations in a way that is not readable by any unauthorized person. And in terms of privacy preservation techniques, we use injection, noise injection, or masking and hiding techniques in order to preserve privacy. The first method of privacy preservation is homomorphic encryption. Homomorphic encryption is a type of encryption where you are doing the same computations and a third party can do your computations for you, but not on the original form of the data, on an encrypted form of data. This means that if you use a third party company for your ML computations, you can totally give your data to them in a transformed form, which is using homomorphic encryption, and they can do the computations for you and give you back the results. And then you can decrypt the data and get the plain text or cipher text as you want. The second method is secure multi-party computation. In case of this technique, we have multiple parties who all contribute into building an ML model or a function or to compute a value for a function. But the thing is that the input is distributed between all of these parties. So as you see, part of input A1 is here, A2 is here, A3 is here, and so on. So this way, if an adversary wants to attack the outputs, they need to be able to hack into all of the parties in order to make sense of the data. Otherwise, getting information from one of the parties is meaningless. Another technique is federated machine learning. Federated machine learning is very similar to the previous method. However, the difference is that we have multiple sets of users, and these users all contribute into building a classifier. The way that we handle the creation of the outputs is that we assign some sort of weights and locally update the average of classifiers by assigning weights to each of the trained models. The most important technique to preserve privacy is differential privacy. Basically, differential privacy means that it's a noise injection technique so that the output of your model shouldn't reveal any information about the input of the model. Whether your model is trained using the data of this person or the data of this person shouldn't be revealed in the outcome of your algorithm. It can be non-interactive or interactive. For example, in case of non-interactive, we pre-compute the stats, the amount of noise, the amount of relationship between the input features, and then inject noise to that. In case of interactive, the amount of noise is injected based on user's requests. What does differential privacy mean? It means that if person i change their information from x i to x prime i, then the probability of their output should not change by that much. What does it mean again? It means that as a person, if one of my features is changing, and as a person who is contributing into training a model, that change shouldn't cause a huge change in the output. Because if that causes that huge change, it will reveal some information about this instance or observance as an input. For example, if I change the color of my hair from brown to green, and that has a huge effect on the output of the model, then just by observing that change, an adversary will know that I changed the color of my hair. So using differential privacy, we want to make that change as minimal as possible. And how do we do that? By computing what is the effect of that change and making that as a noise input into our model. So looking again at the differential privacy, you see here that we have two datasets. We call these neighbor datasets, meaning that all of the features in both datasets are actually equal, except that one feature. In here, it is x3 and x3 prime. We want to make sure that when we change the input data from x3 to x3 prime, the output that is generated by the model is changing minimally. So as you see, the probability of generating a specific output in the first case to the second case is minimal, or epsilon. The next method is called PATE, private aggregation of teacher ensembles. In this case, we have several classifiers that use some data use some features in order to train a model. Then we have two sets of voters. It's like an inversion programming type of technique. These voters choose the majority voting generated by all of these classifiers. And if we have equal amount of votes, for example, in this case, we have two cancers and two healthy results, then we will calculate a Gaussian noise to inject to the voters data. So basically, we again go back into our training data, and we take a look and see that how much noise does each classifier needs to be injected into. Then again, we look at we inject that noise and we get the results off of this chart based on the results. So based on the amount of noise injected, the model might generate the value cancer or might generate the value healthy. The important thing is that whatever the output is, for example, in this case, it is cancer, the adversary cannot correlate it back to the data that is trained in order to get the value of cancer here. Because even if we don't have we have the value healthy here by the noise injection, these might also generate the value cancer. So we don't really know is James Smith the cause of generating the result cancer or if someone similar to James Smith in this training data set is contributing into that result. Here is the comparison of the defense methods in terms of what they're emphasizing on, what are they trying to protect, whether it is the data owner or the model protection, and their use cases in different applications. Most importantly, most of these methods are really effective when they're used in combination with each other, one from each category. For example, homomorphic encryption and differential privacy or SMP and differential privacy, PATE and federated lightning plus homomorphic encryption. Here are the practical methods for privacy preservation. They are very easy to use. There are already packages that are generated for that like TensorFlow or PyTorch or IBM privacy package. We are going to show an example of how to use one of these packages in a real world sample. In this demo, we're going to show how you're going to implement a solution from an adversary point of view, and then from a defense point of view, how you can easily apply differentiable private method into your model. So here in the references, you'll see the data that we use, which is a data source on Kaggle of the purchase dataset. This contains information that you can use to create labels, such as a binary classifier. In some cases, it could be like fraud, or if it's fraudulent or not, or it could be abusive or not. Then we can also create a set of multi-class labels itself, such as the class for a customer, or it could be the category of a merchant, or any other multi-class that you can think of relevant to your use case. And you can see that we're also going to show you the TensorFlow privacy method as well. So first, we're going to look at how a shadow network is created. It's really easy. First, you create the set of shadow networks. Then you output the predictions from the shadow networks, feed that into a TAC network, and then train a target network that is similar to what you think the production environment would look like, and then see whether the prediction from that target network is predicted as an in-member or out-member from the TAC network. If we look at the shadow network, all you have to do is create a very simple for loop. So if you go down, you can see that all you have to do is create n number of shadow networks. They could be similar to each other, or they could be of different structure. But for our use case, we're going to create one type of TensorFlow model, a very simple one with an initial density of 64 nodes, and then dropout, and then another dense layer of 24 nodes, and then finally the softmax for a prediction of the multi-class or binary class. Each of these n shadow networks will have the same networks, and it'll produce a prediction of how likely it is to be in a certain class or not. If we look at the TAC network, this is where then you're feeding the predictions of the shadow networks. We're then training a very fine model to distinguish the threshold of whether a particular observation is within a network or not. So this is very simple. You're training samples from the shadow network. You're going to identify if it came from that particular shadow network or not, and that would be pretty much your label. If we go to the target DP, this will reveal how easy it is to implement a differential private method for your defense mechanism if you already have a deep learning TensorFlow model in production. So all you really have to do is import the TensorFlow privacy library and then update your optimizer. So for in cell 11, you can see that the model is the same, but in cell 12, you can see that the only thing that changes is the optimizer, which is a DP gradient descent constant optimizer with four inputs that you would have to fine-tune. Dial-to-norm clip, noise multiplier, non-micro-batches, and learning rate. And then you have your differential private method. And then if we go back to the first one, here we show results of the different number of shadow networks with the different number of classes, 10 classes, one-inch classes, and the binary two classes, and whether it was differential private or not. And you can see the different types of accuracy, test accuracy, and then the recall and precision rate as well. This is just a summary. If you're interested, you can pause here and look at it for further details. Okay, so this is the end of our presentation. Thank you so much for your time. If you have questions, please reach out to our emails included in the slide deck."
}