title
John Hopfield: Physics View of the Mind and Neurobiology | Lex Fridman Podcast #76

description
John Hopfield is professor at Princeton, whose life's work weaved beautifully through biology, chemistry, neuroscience, and physics. Most crucially, he saw the messy world of biology through the piercing eyes of a physicist. He is perhaps best known for his work on associate neural networks, now known as Hopfield networks that were one of the early ideas that catalyzed the development of the modern field of deep learning. EPISODE LINKS: Now What? article: http://bit.ly/3843LeU John wikipedia: https://en.wikipedia.org/wiki/John_Hopfield Books mentioned: - Einstein's Dreams: https://amzn.to/2PBa96X - Mind is Flat: https://amzn.to/2I3YB84 This episode is presented by Cash App. Download it & use code "LexPodcast": Cash App (App Store): https://apple.co/2sPrUHe Cash App (Google Play): https://bit.ly/2MlvP5w PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ Full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 OUTLINE: 0:00 - Introduction 2:35 - Difference between biological and artificial neural networks 8:49 - Adaptation 13:45 - Physics view of the mind 23:03 - Hopfield networks and associative memory 35:22 - Boltzmann machines 37:29 - Learning 39:53 - Consciousness 48:45 - Attractor networks and dynamical systems 53:14 - How do we build intelligent systems? 57:11 - Deep thinking as the way to arrive at breakthroughs 59:12 - Brain-computer interfaces 1:06:10 - Mortality 1:08:12 - Meaning of life CONNECT: - Subscribe to this YouTube channel - Twitter: https://twitter.com/lexfridman - LinkedIn: https://www.linkedin.com/in/lexfridman - Facebook: https://www.facebook.com/LexFridmanPage - Instagram: https://www.instagram.com/lexfridman - Medium: https://medium.com/@lexfridman - Support on Patreon: https://www.patreon.com/lexfridman

detail
{'title': 'John Hopfield: Physics View of the Mind and Neurobiology | Lex Fridman Podcast #76', 'heatmap': [{'end': 1268.514, 'start': 1219.638, 'weight': 0.705}, {'end': 3106.131, 'start': 3053.094, 'weight': 1}], 'summary': "John hopfield's interdisciplinary work in physics, biology, and neuroscience has significantly impacted deep learning and machine learning, while also exploring the potential breakthroughs in understanding the mind through physics and neural networks, as well as the evolution of ai and neurobiology. the discussion also includes consciousness, memory formation, brain-computer interfaces, and embracing chaos in neural engineering, addressing various aspects of cognition and the interconnectedness of biological systems.", 'chapters': [{'end': 138.879, 'segs': [{'end': 74.564, 'src': 'embed', 'start': 25.005, 'weight': 0, 'content': [{'end': 30.011, 'text': 'that were one of the early ideas that catalyzed the development of the modern field of deep learning.', 'start': 25.005, 'duration': 5.006}, {'end': 35.32, 'text': 'As his 2019 Franklin Medal in Physics award states,', 'start': 31.437, 'duration': 3.883}, {'end': 42.685, 'text': 'he applied concepts of theoretical physics to provide new insights on important biological questions in a variety of areas,', 'start': 35.32, 'duration': 7.365}, {'end': 47.309, 'text': 'including genetics and neuroscience, with significant impact on machine learning.', 'start': 42.685, 'duration': 4.624}, {'end': 52.113, 'text': 'And, as John says in his 2018 article titled Now, What?', 'start': 48.049, 'duration': 4.064}, {'end': 58.976, 'text': 'His accomplishments have often come about by asking that very question now what?', 'start': 52.853, 'duration': 6.123}, {'end': 62.678, 'text': 'And often responding by a major change of direction.', 'start': 59.757, 'duration': 2.921}, {'end': 66.6, 'text': 'This is the Artificial Intelligence Podcast.', 'start': 64.138, 'duration': 2.462}, {'end': 74.564, 'text': 'If you enjoy it, subscribe on YouTube, give it five stars on Apple Podcasts, support it on Patreon or simply connect with me on Twitter.', 'start': 67, 'duration': 7.564}], 'summary': 'Physicist applies theoretical physics to impact genetics, neuroscience, and machine learning.', 'duration': 49.559, 'max_score': 25.005, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE25005.jpg'}], 'start': 0.049, 'title': "John hopfield's interdisciplinary work", 'summary': "Delves into john hopfield's interdisciplinary work in physics, biology, and neuroscience, emphasizing his impact on deep learning and his innovative application of theoretical physics concepts to genetics and neuroscience, with a significant influence on machine learning. it also highlights his approach of asking 'now what?' to drive major changes and his feature on the artificial intelligence podcast.", 'chapters': [{'end': 138.879, 'start': 0.049, 'title': 'John hopfield: physics, biology, and the birth of deep learning', 'summary': "Explores john hopfield's interdisciplinary work in physics, biology, and neuroscience, highlighting his contributions to the development of deep learning and his application of theoretical physics concepts to genetics and neuroscience, with a significant impact on machine learning. it also touches on his approach of asking 'now what?' to bring about major changes and his appearance on the artificial intelligence podcast.", 'duration': 138.83, 'highlights': ["John Hopfield's interdisciplinary work in physics, biology, and neuroscience, particularly his contributions to the development of deep learning and the application of theoretical physics concepts to genetics and neuroscience, with a significant impact on machine learning.", "His approach of asking 'now what?' to bring about major changes in his accomplishments, as highlighted in his 2018 article titled Now, What?", 'His 2019 Franklin Medal in Physics award, which recognized his application of theoretical physics concepts to provide new insights on important biological questions in a variety of areas, including genetics and neuroscience, with a significant impact on machine learning.', "The mention of John Hopfield's appearance on the Artificial Intelligence Podcast and the various ways to support the podcast, including subscribing on YouTube, giving it five stars on Apple Podcasts, and connecting on Twitter."]}], 'duration': 138.83, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE49.jpg', 'highlights': ["John Hopfield's interdisciplinary work in physics, biology, and neuroscience significantly impacted deep learning and machine learning.", "His approach of asking 'now what?' drove major changes in his accomplishments, as highlighted in his 2018 article titled Now, What?", 'Receiving the 2019 Franklin Medal in Physics for applying theoretical physics concepts to provide new insights on important biological questions, impacting genetics, neuroscience, and machine learning.', "John Hopfield's appearance on the Artificial Intelligence Podcast and ways to support the podcast."]}, {'end': 824.565, 'segs': [{'end': 231.97, 'src': 'embed', 'start': 200.395, 'weight': 1, 'content': [{'end': 209.121, 'text': 'And so you expect, in neurobiology, for evolution to have captured all kinds of possibilities of getting neurons,', 'start': 200.395, 'duration': 8.726}, {'end': 211.543, 'text': 'of how you get neurons to do things for you.', 'start': 209.121, 'duration': 2.422}, {'end': 218.401, 'text': 'and that aspect has been completely suppressed in artificial neural networks.', 'start': 213.237, 'duration': 5.164}, {'end': 227.967, 'text': 'Do the glitches become features in the biological neural network? They can.', 'start': 220.282, 'duration': 7.685}, {'end': 231.97, 'text': 'Look, let me take one of the things that I used to do research on.', 'start': 227.987, 'duration': 3.983}], 'summary': 'Neurobiology suggests diverse neuron possibilities, unlike in artificial neural networks.', 'duration': 31.575, 'max_score': 200.395, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE200395.jpg'}, {'end': 295.896, 'src': 'embed', 'start': 265.926, 'weight': 6, 'content': [{'end': 267.506, 'text': "They don't walk in lock step.", 'start': 265.926, 'duration': 1.58}, {'end': 270.468, 'text': 'But they all walk at about the same frequency.', 'start': 268.307, 'duration': 2.161}, {'end': 277.87, 'text': 'And the bridge could sway at that frequency, and the slight sway made pedestrians tend a little bit to lock into step.', 'start': 270.488, 'duration': 7.382}, {'end': 283.873, 'text': 'And after a while, the bridge was oscillating back and forth, and the pedestrians were walking in step to it.', 'start': 277.95, 'duration': 5.923}, {'end': 286.134, 'text': 'And you could see it in the movies made out of the bridge.', 'start': 283.893, 'duration': 2.241}, {'end': 289.495, 'text': 'And the engineers made a simple-minded mistake.', 'start': 286.994, 'duration': 2.501}, {'end': 295.896, 'text': "They assume when you walk, it's step, step, step, and it's back and forth motion.", 'start': 290.592, 'duration': 5.304}], 'summary': 'Pedestrians unintentionally synchronized with swaying bridge.', 'duration': 29.97, 'max_score': 265.926, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE265926.jpg'}, {'end': 394.844, 'src': 'embed', 'start': 339.006, 'weight': 0, 'content': [{'end': 342.909, 'text': 'Well, if they fire together, you can be sure that other cells are going to notice it.', 'start': 339.006, 'duration': 3.903}, {'end': 348.492, 'text': 'So you can make a computational feature out of this in an evolving brain.', 'start': 343.369, 'duration': 5.123}, {'end': 356.695, 'text': "Most artificial neural networks don't even have action potentials, let alone have the possibility for synchronizing them.", 'start': 350.37, 'duration': 6.325}, {'end': 362.099, 'text': 'And you mentioned the evolutionary process.', 'start': 359.277, 'duration': 2.822}, {'end': 369.865, 'text': 'The evolutionary process that builds on top of biological systems leverages that..', 'start': 362.119, 'duration': 7.746}, {'end': 374.902, 'text': 'the weird mess of it somehow.', 'start': 373.48, 'duration': 1.422}, {'end': 383.651, 'text': 'So how do you make sense of that ability to leverage all the different kinds of complexities in the biological brain??', 'start': 375.362, 'duration': 8.289}, {'end': 394.844, 'text': "Well, look, in the biological molecule level, You'd have a piece of DNA which would encode for a particular protein.", 'start': 384.673, 'duration': 10.171}], 'summary': 'Artificial neural networks lack action potentials, hindering synchronizing; leveraging biological complexities for computational features.', 'duration': 55.838, 'max_score': 339.006, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE339006.jpg'}, {'end': 720.999, 'src': 'embed', 'start': 692.402, 'weight': 2, 'content': [{'end': 698.387, 'text': 'the kind of math is doing enables you to solve problems of a very different kind.', 'start': 692.402, 'duration': 5.985}, {'end': 700.749, 'text': "that's right, that's right.", 'start': 698.387, 'duration': 2.362}, {'end': 710.097, 'text': 'so you mentioned two kinds of adaptation the evolutionary adaptation at the and the adaptation or learning at the scale of a single human life.', 'start': 700.749, 'duration': 9.348}, {'end': 710.718, 'text': 'which do you?', 'start': 710.097, 'duration': 0.621}, {'end': 720.999, 'text': 'Which is particularly beautiful to you and interesting from a research and from just a human perspective?', 'start': 712.977, 'duration': 8.022}], 'summary': 'Math enables solving diverse problems, including evolutionary and human adaptation.', 'duration': 28.597, 'max_score': 692.402, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE692402.jpg'}], 'start': 138.879, 'title': 'Neural networks and evolutionary systems', 'summary': "Discusses the difference between biological and artificial neural networks, the potential of glitches becoming features in biological neural networks, the synchronization of oscillating systems, and the brain's adaptive capabilities.", 'chapters': [{'end': 231.97, 'start': 138.879, 'title': 'Neural networks: biological vs. artificial', 'summary': 'Discusses the difference between biological and artificial neural networks and highlights the intriguing aspect of evolution shaping neurons, a feature completely suppressed in artificial neural networks, leading to the potential of glitches becoming features in biological neural networks.', 'duration': 93.091, 'highlights': ['The aspect of evolution shaping neurons and capturing all kinds of possibilities in biological neural networks.', 'The potential of glitches becoming features in biological neural networks due to evolutionary processes.', 'The contrast between the properties and evolutionary adaptation of neurons in biological neural networks and the suppression of these aspects in artificial neural networks.']}, {'end': 824.565, 'start': 231.99, 'title': 'Neural networks and evolutionary systems', 'summary': "Discusses the synchronization of oscillating systems using the example of the millennium bridge, the biological basis for synchronization, and the computational and evolutionary aspects of the brain's adaptive capabilities.", 'duration': 592.575, 'highlights': ["The Millennium Bridge's sway caused pedestrians to synchronize their walking, leading to its closure for stiffening. The slight sway of the Millennium Bridge caused pedestrians to synchronize their walking, leading to its closure for two years for stiffening.", 'The biological basis for synchronization is illustrated through the example of nerve cells producing action potentials at the same rate and the potential for computational features in an evolving brain. The biological basis for synchronization is illustrated through nerve cells producing action potentials at the same rate and the potential for computational features in an evolving brain.', 'The evolutionary process leverages duplicating DNA and allowing molecules to drift apart, leading to the improvement of functions and the ability to adapt to different complexities in the biological brain. The evolutionary process leverages duplicating DNA, allowing molecules to drift apart, and improving functions to adapt to different complexities in the biological brain.', 'The discussion of adaptation includes the differences between evolutionary adaptation and learning at the scale of a single human life, with a focus on developmental neurobiology and early stages of human life. The discussion of adaptation includes the differences between evolutionary adaptation and learning at the scale of a single human life, with a focus on developmental neurobiology and early stages of human life.']}], 'duration': 685.686, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE138879.jpg', 'highlights': ['The evolutionary process leverages duplicating DNA, allowing molecules to drift apart, and improving functions to adapt to different complexities in the biological brain.', 'The potential of glitches becoming features in biological neural networks due to evolutionary processes.', 'The discussion of adaptation includes the differences between evolutionary adaptation and learning at the scale of a single human life, with a focus on developmental neurobiology and early stages of human life.', 'The biological basis for synchronization is illustrated through nerve cells producing action potentials at the same rate and the potential for computational features in an evolving brain.', 'The aspect of evolution shaping neurons and capturing all kinds of possibilities in biological neural networks.', 'The contrast between the properties and evolutionary adaptation of neurons in biological neural networks and the suppression of these aspects in artificial neural networks.', "The Millennium Bridge's sway caused pedestrians to synchronize their walking, leading to its closure for two years for stiffening."]}, {'end': 1130.584, 'segs': [{'end': 880.09, 'src': 'embed', 'start': 856.133, 'weight': 0, 'content': [{'end': 866.476, 'text': 'And the way I pick problems is very characteristic of physics and of an intellectual background which is not psychology, which is not chemistry,', 'start': 856.133, 'duration': 10.343}, {'end': 867.416, 'text': 'and so on and so on.', 'start': 866.476, 'duration': 0.94}, {'end': 869.957, 'text': 'Both of your parents are physicists.', 'start': 868.477, 'duration': 1.48}, {'end': 880.09, 'text': 'Both of my parents were physicists, and the real thing I got out of that was a feeling that the world is an understandable place.', 'start': 870.202, 'duration': 9.888}], 'summary': 'Physics background led to understanding the world', 'duration': 23.957, 'max_score': 856.133, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE856133.jpg'}, {'end': 970.156, 'src': 'embed', 'start': 938.291, 'weight': 2, 'content': [{'end': 941.314, 'text': 'I had this nebulous idea of understanding.', 'start': 938.291, 'duration': 3.023}, {'end': 952.211, 'text': 'So if you looked at a situation, you could say, oh, I expect the bull to make that trajectory, or I expect some intuitive notion of understanding.', 'start': 942.287, 'duration': 9.924}, {'end': 960.054, 'text': "And I don't know how to express that very well, and I've never known how to express it well,", 'start': 952.751, 'duration': 7.303}, {'end': 970.156, 'text': 'And you run smack up against it when you look at these simple neural nets, feed-forward neural nets,', 'start': 961.289, 'duration': 8.867}], 'summary': 'Struggling to express intuitive understanding of neural nets.', 'duration': 31.865, 'max_score': 938.291, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE938291.jpg'}, {'end': 1029.964, 'src': 'embed', 'start': 1000.266, 'weight': 1, 'content': [{'end': 1001.828, 'text': 'Could answer that in two ways.', 'start': 1000.266, 'duration': 1.562}, {'end': 1012.152, 'text': 'I think if you look at real systems, Feedback is an essential aspect of how these real systems compute.', 'start': 1001.928, 'duration': 10.224}, {'end': 1020.274, 'text': 'On the other hand, if I have a mathematical system with feedback, I know I can unlayer this and do it.', 'start': 1013.072, 'duration': 7.202}, {'end': 1027.637, 'text': 'But I have an exponential expansion in the amount of stuff I have to build if I can solve the problem that way.', 'start': 1021.415, 'duration': 6.222}, {'end': 1029.964, 'text': 'So feedback is essential.', 'start': 1028.803, 'duration': 1.161}], 'summary': 'Feedback is an essential aspect of real systems, impacting computation and creating an exponential expansion in problem-solving.', 'duration': 29.698, 'max_score': 1000.266, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1000266.jpg'}, {'end': 1130.584, 'src': 'embed', 'start': 1093.398, 'weight': 3, 'content': [{'end': 1099.441, 'text': 'These rhythms are utterly absent from anything which goes on at Google.', 'start': 1093.398, 'duration': 6.043}, {'end': 1106.305, 'text': 'Yeah, but the rhythms..', 'start': 1103.844, 'duration': 2.461}, {'end': 1110.167, 'text': "But the rhythms what? So, well, that's like comparing..", 'start': 1106.305, 'duration': 3.862}, {'end': 1111.268, 'text': "Okay, I'll tell you.", 'start': 1110.167, 'duration': 1.101}, {'end': 1112.768, 'text': "It's like you're comparing..", 'start': 1111.348, 'duration': 1.42}, {'end': 1120.598, 'text': 'the greatest classical musician in the world to a child first learning to play.', 'start': 1115.674, 'duration': 4.924}, {'end': 1123.5, 'text': "But they're still both playing the piano.", 'start': 1121.999, 'duration': 1.501}, {'end': 1128.523, 'text': "I'm asking will it ever go on at Google??", 'start': 1124.68, 'duration': 3.843}, {'end': 1130.584, 'text': 'Do you have a hope??', 'start': 1129.784, 'duration': 0.8}], 'summary': 'Comparison of rhythms at google to a child learning piano.', 'duration': 37.186, 'max_score': 1093.398, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1093398.jpg'}], 'start': 825.306, 'title': 'Understanding the mind through physics', 'summary': 'Explores potential breakthroughs in understanding the mind through physics, neural networks, feedback systems, and the distinction between artificial and biological networks.', 'chapters': [{'end': 1130.584, 'start': 825.306, 'title': 'Understanding the mind through physics', 'summary': 'Discusses the potential breakthrough in understanding the mind, questioning the role of physics, neural networks, feedback systems, and the difference between artificial and biological networks.', 'duration': 305.278, 'highlights': ['Physics as a lens for understanding the world, driven by the belief that the world is an understandable place through experiments and mathematics.', "Challenges in defining 'understanding' in the context of neural networks and the limitations of feed-forward systems in achieving true understanding.", 'The essential role of feedback systems in computation, despite the complexity and exponential expansion in solving problems using feedback.', "Comparison between the brain's electrical rhythms in determining functionality and the absence of such rhythms in artificial systems like Google, highlighting the disparity in understanding between biological and artificial networks."]}], 'duration': 305.278, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE825306.jpg', 'highlights': ['Physics as a lens for understanding the world, driven by experiments and mathematics.', 'The essential role of feedback systems in computation, despite complexity and exponential expansion.', "Challenges in defining 'understanding' in the context of neural networks and limitations of feed-forward systems.", "Comparison between brain's electrical rhythms and absence of such rhythms in artificial systems like Google."]}, {'end': 1660.94, 'segs': [{'end': 1268.514, 'src': 'heatmap', 'start': 1219.638, 'weight': 0.705, 'content': [{'end': 1225.66, 'text': 'Yeah And going back to my brainwaves, as it were.', 'start': 1219.638, 'duration': 6.022}, {'end': 1237.34, 'text': 'Yes From the AI point of view, they would say, ah, maybe these are an epiphenomenon and not important at all.', 'start': 1226, 'duration': 11.34}, {'end': 1249.831, 'text': 'The first car I had, a real wreck of a 1936 Dodge, go above about 45 miles an hour and the wheels would shimmy.', 'start': 1240.483, 'duration': 9.348}, {'end': 1254.475, 'text': 'Good speedometer, that.', 'start': 1252.613, 'duration': 1.862}, {'end': 1259.746, 'text': 'Now, nobody designed the car that way.', 'start': 1256.843, 'duration': 2.903}, {'end': 1261.688, 'text': 'The car is malfunctioning to have that.', 'start': 1259.766, 'duration': 1.922}, {'end': 1268.514, 'text': 'But in biology, if it were useful to know when are you going more than 45 miles an hour,', 'start': 1262.068, 'duration': 6.446}], 'summary': 'Ai considers brainwaves as epiphenomenon, not designed like malfunctioning car. no purpose in biology.', 'duration': 48.876, 'max_score': 1219.638, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1219638.jpg'}, {'end': 1313.273, 'src': 'embed', 'start': 1287.777, 'weight': 0, 'content': [{'end': 1293.501, 'text': 'look the um, how many transistors are there in your laptop these days?', 'start': 1287.777, 'duration': 5.724}, {'end': 1295.643, 'text': "actually, i don't.", 'start': 1293.501, 'duration': 2.142}, {'end': 1296.483, 'text': "i don't know the number.", 'start': 1295.643, 'duration': 0.84}, {'end': 1299.125, 'text': "it's, it's on the scale of 10 to the 10.", 'start': 1296.483, 'duration': 2.642}, {'end': 1300.806, 'text': "i can't remember the number either.", 'start': 1299.125, 'duration': 1.681}, {'end': 1312.072, 'text': 'yeah, and all the transistors are somewhat similar and most physical systems, with that many parts, all of which are similar,', 'start': 1300.806, 'duration': 11.266}, {'end': 1313.273, 'text': 'have collective properties.', 'start': 1312.072, 'duration': 1.201}], 'summary': 'Laptop has around 10 billion transistors, with collective properties.', 'duration': 25.496, 'max_score': 1287.777, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1287777.jpg'}, {'end': 1381.688, 'src': 'embed', 'start': 1345.349, 'weight': 1, 'content': [{'end': 1352.355, 'text': 'We might have to return several times to neurobiology and try to make our transistors more messy.', 'start': 1345.349, 'duration': 7.006}, {'end': 1353.796, 'text': 'Yeah, yeah.', 'start': 1353.035, 'duration': 0.761}, {'end': 1361.041, 'text': 'At the same time, the simple ones will conquer big aspects.', 'start': 1355.259, 'duration': 5.782}, {'end': 1376.406, 'text': 'And I think one of the biggest surprises to me was how well learning systems, which are manifestly non-biological,', 'start': 1363.682, 'duration': 12.724}, {'end': 1381.688, 'text': 'how important they can be actually and how important and how useful they can be in AI.', 'start': 1376.406, 'duration': 5.282}], 'summary': 'Neurobiology-inspired transistors may need multiple revisions; learning systems are crucial for ai.', 'duration': 36.339, 'max_score': 1345.349, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1345349.jpg'}, {'end': 1500.989, 'src': 'embed', 'start': 1471.894, 'weight': 2, 'content': [{'end': 1483.58, 'text': "And it's this ability to link things together, link experiences together, which goes under the general name of associative memory.", 'start': 1471.894, 'duration': 11.686}, {'end': 1492.425, 'text': 'And a large part of intelligent behavior is actually just large associative memories that work, as far as I can see.', 'start': 1484.601, 'duration': 7.824}, {'end': 1498.027, 'text': 'What do you think is the mechanism of how it works in the mind?', 'start': 1493.745, 'duration': 4.282}, {'end': 1500.989, 'text': 'Is it a mystery to you still?', 'start': 1499.548, 'duration': 1.441}], 'summary': 'Intelligent behavior relies on large associative memories in the mind.', 'duration': 29.095, 'max_score': 1471.894, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1471894.jpg'}], 'start': 1131.085, 'title': 'Ai, neurobiology, and associative memory', 'summary': 'Explores the evolution of ai and neurobiology, highlighting the potential for advancements, the scale of physical systems, and the surprising effectiveness of learning systems in ai. it also delves into associative memory, learning mechanisms, and the role of associative artificial networks in intelligent behavior.', 'chapters': [{'end': 1313.273, 'start': 1131.085, 'title': 'The future of ai and neurobiology', 'summary': 'Discusses the evolution of ai and neurobiology, highlighting the potential for generations of advancements, the importance of understanding brain functions, and the massive scale of physical systems like laptops with approximately 10^10 transistors.', 'duration': 182.188, 'highlights': ['Understanding the evolution of AI and neurobiology, with potential for generations of advancements', 'The importance of capturing brain functions for computation, similar to the use of collective properties in physical systems', 'The massive scale of physical systems like laptops with approximately 10^10 transistors']}, {'end': 1470.502, 'start': 1315.475, 'title': 'Neural networks and associative memory', 'summary': 'Discusses the absence of collective properties in artificial neural networks compared to biological systems, emphasizes the surprising effectiveness of learning systems in ai, and explores the concept of associative memory in human cognition, illustrated through the example of identifying a person based on associated traits.', 'duration': 155.027, 'highlights': ['The surprising effectiveness of learning systems in AI, despite being non-biological, has been a significant development, leading to recent advancements in neural networks.', 'The absence of collective properties in artificial neural networks, unlike in biological systems, presents a challenge that may require revisiting neurobiology for further insights.', 'Exploring the concept of associative memory, where a few associated facts enable the recollection of additional information, exemplified by recognizing a person based on specific traits.']}, {'end': 1660.94, 'start': 1471.894, 'title': 'Associative memory and learning', 'summary': 'Discusses the concept of associative memory in intelligent behavior, the mechanism of learning, and the compaction of information into useful chunks, highlighting the role of associative artificial networks and the automatic nature of memory storage in the mind.', 'duration': 189.046, 'highlights': ['The chapter highlights the role of associative memory in intelligent behavior, suggesting that a large part of intelligent behavior is based on associative memories that work effectively.', 'It discusses the mechanism of learning and understanding, emphasizing the need to link things together and learn how to compress information into useful chunks.', 'The chapter also mentions the creation of associative artificial networks to understand learning processes and the compaction of information, providing a physical understanding of learning and cognition.']}], 'duration': 529.855, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1131085.jpg', 'highlights': ['The massive scale of physical systems like laptops with approximately 10^10 transistors', 'The surprising effectiveness of learning systems in AI, despite being non-biological, has been a significant development, leading to recent advancements in neural networks.', 'The chapter highlights the role of associative memory in intelligent behavior, suggesting that a large part of intelligent behavior is based on associative memories that work effectively.']}, {'end': 2525.07, 'segs': [{'end': 1813.082, 'src': 'embed', 'start': 1784.654, 'weight': 0, 'content': [{'end': 1790.396, 'text': "And I can't just get by by saying I'll do the dynamics of activity with a fixed synapses.", 'start': 1784.654, 'duration': 5.742}, {'end': 1798.019, 'text': 'So the synaptic, the dynamics of the synapses is actually fundamental to the whole system.', 'start': 1792.577, 'duration': 5.442}, {'end': 1798.799, 'text': 'Yeah, yeah.', 'start': 1798.199, 'duration': 0.6}, {'end': 1804.091, 'text': "And there's nothing necessarily separating the time scales.", 'start': 1800.006, 'duration': 4.085}, {'end': 1806.454, 'text': 'When the time scales can be separated.', 'start': 1804.772, 'duration': 1.682}, {'end': 1813.082, 'text': "it's neat from the physicist's or the mathematician's point of view, but it's not necessarily true in neurobiology.", 'start': 1806.454, 'duration': 6.628}], 'summary': 'Synaptic dynamics are fundamental to the whole system, with no strict time scale separation in neurobiology.', 'duration': 28.428, 'max_score': 1784.654, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1784654.jpg'}, {'end': 1925.061, 'src': 'embed', 'start': 1895.736, 'weight': 2, 'content': [{'end': 1902.102, 'text': "which are both true and for which you don't need all the molecular details of the molecules colliding.", 'start': 1895.736, 'duration': 6.366}, {'end': 1907.508, 'text': "That's what I mean from the roots of physics by understanding.", 'start': 1903.584, 'duration': 3.924}, {'end': 1911.133, 'text': 'So what did again?', 'start': 1909.212, 'duration': 1.921}, {'end': 1914.775, 'text': 'sorry, but Hopfield Networks help you understand.', 'start': 1911.133, 'duration': 3.642}, {'end': 1919.618, 'text': 'what insight did it give us about memory, about learning?', 'start': 1914.775, 'duration': 4.843}, {'end': 1925.061, 'text': "They didn't give insights about learning.", 'start': 1921.579, 'duration': 3.482}], 'summary': 'Hopfield networks provide insights into memory, not learning.', 'duration': 29.325, 'max_score': 1895.736, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1895736.jpg'}, {'end': 2099.375, 'src': 'embed', 'start': 2066.641, 'weight': 1, 'content': [{'end': 2069.083, 'text': 'It gets back and follows the course of the river.', 'start': 2066.641, 'duration': 2.442}, {'end': 2079.045, 'text': "And that's basically the analog in the physical system which enables you to say oh yes,", 'start': 2069.103, 'duration': 9.942}, {'end': 2089.141, 'text': 'Error-free computation and associative memory are very much like things that I can understand from the point of view of a physical system.', 'start': 2080.675, 'duration': 8.466}, {'end': 2095.565, 'text': 'The physical system can be, under some circumstances, an accurate metaphor.', 'start': 2090.261, 'duration': 5.304}, {'end': 2099.375, 'text': 'not the only metaphor.', 'start': 2098.314, 'duration': 1.061}], 'summary': 'Analog and physical system enable error-free computation and associative memory.', 'duration': 32.734, 'max_score': 2066.641, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2066641.jpg'}, {'end': 2191.521, 'src': 'embed', 'start': 2165.913, 'weight': 3, 'content': [{'end': 2176.996, 'text': 'It always intrigued me that one of the most long-lived of the learning systems is the Boltzmann machine, which is intrinsically a feedback network.', 'start': 2165.913, 'duration': 11.083}, {'end': 2186.579, 'text': 'And it was the brilliance of Hinton and Zanowski to understand how to do learning in that.', 'start': 2179.157, 'duration': 7.422}, {'end': 2191.521, 'text': "And it's still a useful way to understand, learning and understand,", 'start': 2188.28, 'duration': 3.241}], 'summary': 'The boltzmann machine, a long-lived learning system, was developed by hinton and zanowski for understanding and implementing learning processes.', 'duration': 25.608, 'max_score': 2165.913, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2165913.jpg'}, {'end': 2314.529, 'src': 'embed', 'start': 2281.74, 'weight': 5, 'content': [{'end': 2284.783, 'text': 'Or is it something fundamentally different going on in the brain??', 'start': 2281.74, 'duration': 3.043}, {'end': 2299.777, 'text': "I don't think the brain is as deep as the deepest networks go, the deepest computer science networks.", 'start': 2290.33, 'duration': 9.447}, {'end': 2314.529, 'text': "And I do wonder whether part of that depth of the computer science networks is necessitated by the fact that the only learning that's easily done on a machine is feed forward.", 'start': 2302.219, 'duration': 12.31}], 'summary': 'Brain may not be as deep as computer networks, machine learning is feed forward', 'duration': 32.789, 'max_score': 2281.74, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2281740.jpg'}, {'end': 2465.673, 'src': 'embed', 'start': 2430.19, 'weight': 4, 'content': [{'end': 2438.99, 'text': 'Well, I asked Marvin Minsky, his view on consciousness, and Marvin said, Consciousness is basically overrated.', 'start': 2430.19, 'duration': 8.8}, {'end': 2442.053, 'text': 'It may be an epiphenomenon.', 'start': 2439.01, 'duration': 3.043}, {'end': 2451.301, 'text': 'After all, all the things your brain does, which are actually hard computations, you do non-consciously.', 'start': 2442.894, 'duration': 8.407}, {'end': 2462.352, 'text': "And there's so much evidence that even the simple things you do can make decisions.", 'start': 2455.725, 'duration': 6.627}, {'end': 2465.673, 'text': 'you can make committed decisions about them.', 'start': 2462.352, 'duration': 3.321}], 'summary': 'Marvin minsky views consciousness as overrated and possibly an epiphenomenon, as the brain performs hard computations non-consciously with evidence supporting the ability to make decisions.', 'duration': 35.483, 'max_score': 2430.19, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2430190.jpg'}], 'start': 1661.961, 'title': 'Synaptic dynamics and neural network evolution', 'summary': 'Delves into understanding synapse dynamics in neurobiology, exploring the interplay between physics and biology, error correction in computation, and limitations of hopfield networks, along with the evolution of neural networks, impact of feedback networks on learning, and exploration of consciousness through influential figures like jeff hinton and marvin minsky.', 'chapters': [{'end': 2119.917, 'start': 1661.961, 'title': 'Understanding the dynamics of synapses', 'summary': 'Delves into the complexity of understanding the dynamics of synapses in neurobiology, detailing the interplay between physics and biology, the concept of error correction in computation, and the limitations of hopfield networks in describing the learning process.', 'duration': 457.956, 'highlights': ['The dynamics of synapses in neurobiology and the interplay between physics and biology The chapter extensively discusses the interplay between physics and biology, emphasizing the fundamental importance of understanding the dynamics of synapses in neurobiology.', 'Concept of error correction in computation and its analogy to physical systems The concept of error correction in computation is explained using the analogy of a physical system, highlighting its importance in achieving error-free computation and associative memory.', 'Limitations of Hopfield networks in describing the learning process The chapter reveals the limitations of Hopfield networks in providing a reasonable description of the learning process, focusing more on how learned things can be expressed rather than the actual learning process.']}, {'end': 2525.07, 'start': 2122.742, 'title': 'Neural network evolution', 'summary': 'Discusses the evolution of neural networks and the impact of feedback networks on computational and physical learning, as well as the exploration of consciousness through the perspective of influential figures in the field, such as jeff hinton and marvin minsky.', 'duration': 402.328, 'highlights': ['The impact of feedback networks on computational and physical learning is discussed, specifically the role of Boltzmann machines in understanding learning and its relation to feed-forward systems. Discussion of the impact of feedback networks on computational and physical learning, including the role of Boltzmann machines and their relation to feed-forward systems.', 'Exploration of consciousness through the perspectives of influential figures like Marvin Minsky, who suggests that consciousness may be an overrated epiphenomenon, with the real computational activities occurring at the non-conscious level. Exploring consciousness through the perspectives of influential figures like Marvin Minsky, who suggests that consciousness may be overrated and that real computational activities occur at the non-conscious level.', 'Comparative analysis between the depth of computer science networks and the complexity of the brain, questioning the necessity of feedback in capturing the dynamics of biological learning processes. Comparative analysis between the depth of computer science networks and the complexity of the brain, questioning the necessity of feedback in capturing the dynamics of biological learning processes.']}], 'duration': 863.109, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE1661961.jpg', 'highlights': ['The dynamics of synapses in neurobiology and the interplay between physics and biology are extensively discussed.', 'The concept of error correction in computation is explained using the analogy of a physical system.', 'The chapter reveals the limitations of Hopfield networks in providing a reasonable description of the learning process.', 'The impact of feedback networks on computational and physical learning is discussed, including the role of Boltzmann machines.', 'Exploring consciousness through the perspectives of influential figures like Marvin Minsky.', 'Comparative analysis between the depth of computer science networks and the complexity of the brain.']}, {'end': 3351.097, 'segs': [{'end': 2558.433, 'src': 'embed', 'start': 2525.07, 'weight': 0, 'content': [{'end': 2534.171, 'text': 'he would probably have said consciousness is your effort to explain to yourself that which you have already done.', 'start': 2525.07, 'duration': 9.101}, {'end': 2542.679, 'text': "Yeah, it's the weaving of the narrative around the things that have already been computed for you.", 'start': 2536.854, 'duration': 5.825}, {'end': 2544.26, 'text': "That's right.", 'start': 2543.76, 'duration': 0.5}, {'end': 2549.825, 'text': 'And so much of what we do for our memories of events.', 'start': 2544.34, 'duration': 5.485}, {'end': 2558.433, 'text': "for example, if there's some traumatic event you witness, you will have a few facts about it, correctly done.", 'start': 2549.825, 'duration': 8.608}], 'summary': 'Consciousness is the weaving of narrative around computed events, aiding memory recall.', 'duration': 33.363, 'max_score': 2525.07, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2525070.jpg'}, {'end': 2663.732, 'src': 'embed', 'start': 2629.44, 'weight': 3, 'content': [{'end': 2642.189, 'text': 'And long afterward, some of the tapes, the secret tapes as it were, from which Jane was recalling these conversations were published.', 'start': 2629.44, 'duration': 12.749}, {'end': 2646.748, 'text': 'And one found out that John Dean had a good but not exceptional memory.', 'start': 2643.246, 'duration': 3.502}, {'end': 2654.811, 'text': 'What he had was an ability to paint vividly and in some sense accurately the tone of what was going on.', 'start': 2647.148, 'duration': 7.663}, {'end': 2663.732, 'text': "By the way, that's a beautiful description of consciousness.", 'start': 2656.926, 'duration': 6.806}], 'summary': 'John dean had a good but not exceptional memory, excelled in painting vivid and accurate tones of conversations.', 'duration': 34.292, 'max_score': 2629.44, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2629440.jpg'}, {'end': 2838.367, 'src': 'embed', 'start': 2813.956, 'weight': 1, 'content': [{'end': 2821.081, 'text': "there's a point where what I know about physics is so different from any neurobiologist that I have something that I might be able to contribute.", 'start': 2813.956, 'duration': 7.125}, {'end': 2827.126, 'text': "And right now, there's no way to grasp at consciousness from a physics perspective.", 'start': 2821.782, 'duration': 5.344}, {'end': 2829.468, 'text': "From my point of view, that's correct.", 'start': 2827.727, 'duration': 1.741}, {'end': 2832.291, 'text': 'And of course, people..', 'start': 2831.55, 'duration': 0.741}, {'end': 2838.367, 'text': 'Physicists, like everybody else, think very muddily about things.', 'start': 2834.124, 'duration': 4.243}], 'summary': 'Physics offers unique perspective on consciousness, lacking current grasp.', 'duration': 24.411, 'max_score': 2813.956, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2813956.jpg'}, {'end': 2993.53, 'src': 'embed', 'start': 2957.878, 'weight': 2, 'content': [{'end': 2967.484, 'text': 'Right And an attractor network is simply a network where there is a line and other lines converge on it in time.', 'start': 2957.878, 'duration': 9.606}, {'end': 2970.406, 'text': "That's the essence of an attractor network.", 'start': 2968.505, 'duration': 1.901}, {'end': 2974.596, 'text': 'In a highly dimensional space.', 'start': 2971.252, 'duration': 3.344}, {'end': 2985.109, 'text': 'And the easiest way to get that is to do it in a highly dimensional space, where some of the dimensions provide the dissipation, which..', 'start': 2974.777, 'duration': 10.332}, {'end': 2993.53, 'text': "If I have a physical system, trajectories can't contract everywhere.", 'start': 2987.566, 'duration': 5.964}], 'summary': 'An attractor network involves convergence of lines in a highly dimensional space.', 'duration': 35.652, 'max_score': 2957.878, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2957878.jpg'}, {'end': 3106.131, 'src': 'heatmap', 'start': 3053.094, 'weight': 1, 'content': [{'end': 3058.518, 'text': 'You have some defined pathways which are allowed and onto which you will converge.', 'start': 3053.094, 'duration': 5.424}, {'end': 3063.722, 'text': "And that's the way you make a stable computer, and that's the way you make a stable behavior.", 'start': 3060.119, 'duration': 3.603}, {'end': 3075.305, 'text': 'So, in general, looking at the physics of the emergent stability in networks,', 'start': 3066.354, 'duration': 8.951}, {'end': 3084.235, 'text': 'what are some interesting insights from studying the dynamics of such high-dimensional systems?', 'start': 3075.305, 'duration': 8.93}, {'end': 3089.845, 'text': 'Most dynamical systems, most driven dynamical systems.', 'start': 3085.003, 'duration': 4.842}, {'end': 3097.388, 'text': "by driven they're coupled somehow to an energy source and so their dynamics keeps going because it's coupling to the energy source.", 'start': 3089.845, 'duration': 7.543}, {'end': 3106.131, 'text': "Most of them, it's very difficult to understand at all what the dynamical behavior is going to be.", 'start': 3100.109, 'duration': 6.022}], 'summary': 'Studying dynamics of high-dimensional systems reveals insights on emergent stability in networks and driven dynamical systems.', 'duration': 53.037, 'max_score': 3053.094, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3053094.jpg'}], 'start': 2525.07, 'title': 'Consciousness and memory formation', 'summary': "Delves into consciousness as the narrative builder around facts, demonstrated through john dean's memory, highlighting memory's subjective nature. it also explores the significance of consciousness in cognition, the complexities in understanding it from physics and neurobiology, and the constraints of basic neural networks in thought processing.", 'chapters': [{'end': 2663.732, 'start': 2525.07, 'title': 'Consciousness and memory formation', 'summary': "Discusses how consciousness involves weaving a narrative around computed facts, illustrated by the example of john dean's vivid yet inaccurate recollection of events, revealing the subjective nature of memory formation.", 'duration': 138.662, 'highlights': ['Consciousness involves weaving a narrative around computed facts, leading to subjective memory formation.', "John Dean's vivid yet inaccurate recollection of events during the Watergate era exemplifies the subjective nature of memory formation.", "The example of John Dean's testimony highlights the subjective nature of memory formation, with his ability to vividly portray events despite not having exceptional memory accuracy.", 'Consciousness is described as the weaving of narrative around computed facts, leading to the creation of a subjective memory that may not be entirely accurate.']}, {'end': 3351.097, 'start': 2663.752, 'title': 'Importance of consciousness in cognition', 'summary': 'Discusses the importance of consciousness in cognition, the challenges in understanding consciousness from physics and neurobiology perspectives, and the limitations of simple neural networks in achieving thought.', 'duration': 687.345, 'highlights': ['The importance of consciousness in cognition and its fundamental role in intelligence is discussed, drawing parallels with the understanding of genetics in biology. The chapter highlights the fundamental role of consciousness in intelligence and draws parallels with the understanding of genetics in biology, emphasizing the deep problem of inheritance and the challenges in comprehending consciousness from a physics perspective.', 'The challenges in understanding consciousness from physics and neurobiology perspectives are discussed, along with the limitations of simple neural networks in achieving thought. It discusses the challenges in comprehending consciousness from physics and neurobiology perspectives and emphasizes the limitations of simple neural networks in achieving thought, particularly in mental exploration and creative elements.', 'The dynamics of high-dimensional systems and the concept of attractor networks are explained, highlighting the behavior of stable pathways and the emergence of stability in networks. The chapter explains the dynamics of high-dimensional systems and the concept of attractor networks, emphasizing the behavior of stable pathways and the emergence of stability in networks, providing insights into the dynamics of driven systems.']}], 'duration': 826.027, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE2525070.jpg', 'highlights': ['Consciousness involves weaving a narrative around computed facts, leading to subjective memory formation.', 'The challenges in comprehending consciousness from physics and neurobiology perspectives and emphasizes the limitations of simple neural networks in achieving thought, particularly in mental exploration and creative elements.', 'The chapter explains the dynamics of high-dimensional systems and the concept of attractor networks, emphasizing the behavior of stable pathways and the emergence of stability in networks, providing insights into the dynamics of driven systems.', "John Dean's vivid yet inaccurate recollection of events during the Watergate era exemplifies the subjective nature of memory formation."]}, {'end': 3694.011, 'segs': [{'end': 3382.093, 'src': 'embed', 'start': 3352.837, 'weight': 0, 'content': [{'end': 3360.123, 'text': 'And so the artificial neural net world is always very much, I have a population of examples.', 'start': 3352.837, 'duration': 7.286}, {'end': 3364.567, 'text': 'The test set must be drawn from the equivalent population.', 'start': 3361.004, 'duration': 3.563}, {'end': 3375.356, 'text': "If the test set has examples which are from a population which is completely different, there's no way that you could expect to get the answer right.", 'start': 3364.587, 'duration': 10.769}, {'end': 3380.34, 'text': 'Yeah, what they call outside the distribution.', 'start': 3376.497, 'duration': 3.843}, {'end': 3382.093, 'text': "That's right, that's right.", 'start': 3380.953, 'duration': 1.14}], 'summary': 'Artificial neural net must have test set drawn from equivalent population to expect accurate results.', 'duration': 29.256, 'max_score': 3352.837, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3352837.jpg'}, {'end': 3444.838, 'src': 'embed', 'start': 3413.571, 'weight': 2, 'content': [{'end': 3417.915, 'text': "The population of the training set isn't just sort of this set of examples.", 'start': 3413.571, 'duration': 4.344}, {'end': 3422.519, 'text': "There's more to it than that.", 'start': 3421.298, 'duration': 1.221}, {'end': 3429.582, 'text': 'And it gets back to my question of, what is it to understand something? Yeah.', 'start': 3423.575, 'duration': 6.007}, {'end': 3440.455, 'text': "You know, in a small tangent, you've talked about the value of thinking of deductive reasoning in science versus large data collection.", 'start': 3432.005, 'duration': 8.45}, {'end': 3444.838, 'text': 'sort of thinking about the problem.', 'start': 3442.677, 'duration': 2.161}], 'summary': 'Training set population goes beyond examples, exploring understanding and deductive reasoning in data science.', 'duration': 31.267, 'max_score': 3413.571, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3413571.jpg'}, {'end': 3645.493, 'src': 'embed', 'start': 3588.979, 'weight': 3, 'content': [{'end': 3600.872, 'text': "So in the case of Neuralink, they're doing a thousand plus connections where they're able to do two-way activate and read spikes, neural spikes.", 'start': 3588.979, 'duration': 11.893}, {'end': 3618.632, 'text': 'Do you have hope for that kind of computer brain interaction in the near or maybe even far future? of being able to expand the ability of the mind of cognition or understand the mind?', 'start': 3601.433, 'duration': 17.199}, {'end': 3623.054, 'text': "It's interesting watching things go.", 'start': 3618.652, 'duration': 4.402}, {'end': 3630.898, 'text': 'When I first became interested in neurobiology, most of the practitioners thought you would be able to understand neurobiology.', 'start': 3623.774, 'duration': 7.124}, {'end': 3636.531, 'text': 'by techniques which allowed you to record only one cell at a time.', 'start': 3631.65, 'duration': 4.881}, {'end': 3637.231, 'text': 'One cell, yeah.', 'start': 3636.551, 'duration': 0.68}, {'end': 3645.493, 'text': 'People like David Hubel very strongly reflected that point of view.', 'start': 3638.611, 'duration': 6.882}], 'summary': 'Neuralink aims to achieve 1000+ connections for two-way neural spike activation, potentially expanding cognition in the future.', 'duration': 56.514, 'max_score': 3588.979, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3588979.jpg'}], 'start': 3352.837, 'title': 'Test set population and brain-computer interfaces', 'summary': 'Emphasizes the importance of matching test set with training set population for accurate results in neural nets. it also explores the value of deductive reasoning in scientific processes and the potential of brain-computer interfaces for expanding cognitive abilities.', 'chapters': [{'end': 3413.191, 'start': 3352.837, 'title': 'Neural net test set population', 'summary': 'Discusses the importance of the test set being drawn from the equivalent population as the training set to ensure accurate results, highlighting the concept of being outside the distribution and the impact on neural net performance.', 'duration': 60.354, 'highlights': ['The test set must be drawn from the equivalent population to the training set to expect accurate results.', 'Examples in the test set from a completely different population will lead to inaccurate answers.', "The concept of being 'outside the distribution' can significantly impact neural net performance.", 'Instances not encountered in the training set may not be recognized by the neural net, highlighting the importance of the test set population.']}, {'end': 3694.011, 'start': 3413.571, 'title': 'Scientific reasoning and brain-computer interfaces', 'summary': 'Discusses the value of deductive reasoning in scientific processes, emphasizing the importance of understanding principles over large data analysis, and explores the potential of brain-computer interfaces for expanding cognitive abilities.', 'duration': 280.44, 'highlights': ['The importance of understanding principles over large data analysis is emphasized, citing examples from cosmology and physics where insight and problem-solving derive from thinking through the problem rather than crunching large data. Examples from cosmology and physics are provided.', 'The discussion on brain-computer interfaces delves into the shift from recording one cell at a time to the need for recording many cells at once to understand the collective operations of the brain. Shift in neurobiological techniques is mentioned.', "The potential of brain-computer interfaces, particularly the work of Neuralink with a focus on a thousand plus connections for two-way activation and reading neural spikes, is explored. Details about Neuralink's work are provided."]}], 'duration': 341.174, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3352837.jpg', 'highlights': ['The test set must be drawn from the equivalent population to the training set to expect accurate results.', "The concept of being 'outside the distribution' can significantly impact neural net performance.", 'The importance of understanding principles over large data analysis is emphasized, citing examples from cosmology and physics where insight and problem-solving derive from thinking through the problem rather than crunching large data.', 'The discussion on brain-computer interfaces delves into the shift from recording one cell at a time to the need for recording many cells at once to understand the collective operations of the brain.', 'The potential of brain-computer interfaces, particularly the work of Neuralink with a focus on a thousand plus connections for two-way activation and reading neural spikes, is explored.']}, {'end': 4359.075, 'segs': [{'end': 3755.76, 'src': 'embed', 'start': 3724.353, 'weight': 4, 'content': [{'end': 3735.096, 'text': 'An engineer would put in six highly accurate stepping motors controlling a limb rather than 100,000 muscle fibers,', 'start': 3724.353, 'duration': 10.743}, {'end': 3737.177, 'text': 'each of which has to be individually controlled.', 'start': 3735.096, 'duration': 2.081}, {'end': 3750.358, 'text': 'And so understanding how to do things in a way which is much more forgiving and much more neural, I think, would benefit the engineering world.', 'start': 3739.558, 'duration': 10.8}, {'end': 3755.76, 'text': 'The engineering world, ah, touch.', 'start': 3753.78, 'duration': 1.98}], 'summary': 'Using 6 motors instead of 100,000 muscle fibers for better control in engineering.', 'duration': 31.407, 'max_score': 3724.353, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3724353.jpg'}, {'end': 4007.234, 'src': 'embed', 'start': 3937.066, 'weight': 2, 'content': [{'end': 3938.427, 'text': 'You have a sense that..', 'start': 3937.066, 'duration': 1.361}, {'end': 3941.968, 'text': 'Yeah, but I can only dream physics dreams.', 'start': 3938.427, 'duration': 3.541}, {'end': 3943.308, 'text': 'Physics dreams.', 'start': 3942.348, 'duration': 0.96}, {'end': 3946.829, 'text': "There was an interesting book called Einstein's Dreams,", 'start': 3944.188, 'duration': 2.641}, {'end': 3957.832, 'text': "which alternates between chapters on his life and descriptions of the way time might have been but isn't.", 'start': 3946.829, 'duration': 11.003}, {'end': 3969.132, 'text': 'The linking between these being, of course, ideas that Einstein might have had to think about the essence of time as he was thinking about time.', 'start': 3960.146, 'duration': 8.986}, {'end': 3978.66, 'text': "So, speaking of the essence of time and neurobiology, you're one human, famous, impactful human,", 'start': 3971.354, 'duration': 7.306}, {'end': 3982.363, 'text': 'but just one human with a brain living the human condition.', 'start': 3978.66, 'duration': 3.703}, {'end': 3986.926, 'text': "But you're ultimately mortal, just like all of us.", 'start': 3984.164, 'duration': 2.762}, {'end': 3993.171, 'text': 'Has studying the mind as a mechanism changed the way you think about your own mortality?', 'start': 3987.627, 'duration': 5.544}, {'end': 4007.234, 'text': 'It has really because, particularly as you get older and the body comes apart in various ways,', 'start': 3998.729, 'duration': 8.505}], 'summary': "Discussion on einstein's dreams and mortality in neurobiology.", 'duration': 70.168, 'max_score': 3937.066, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3937066.jpg'}, {'end': 4230.392, 'src': 'embed', 'start': 4203.085, 'weight': 1, 'content': [{'end': 4210.467, 'text': "And you're saying that it's all interconnected in some kind of way that there might not even be an individual.", 'start': 4203.085, 'duration': 7.382}, {'end': 4220.229, 'text': "We're all kind of this complicated mess of biological systems at all different levels, where the human starts and when the human ends is unclear.", 'start': 4210.627, 'duration': 9.602}, {'end': 4222.106, 'text': 'Yeah, yeah.', 'start': 4221.405, 'duration': 0.701}, {'end': 4230.392, 'text': "And we're in neurobiology where the, oh, you say the neocortex is the thinking, but there's lots of things that are done on the spinal cord.", 'start': 4222.146, 'duration': 8.246}], 'summary': 'The interconnected nature of human biology blurs the line between individuality and the role of different biological systems.', 'duration': 27.307, 'max_score': 4203.085, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE4203085.jpg'}, {'end': 4325.71, 'src': 'embed', 'start': 4296.972, 'weight': 0, 'content': [{'end': 4302.835, 'text': 'And now let me leave you with some words of wisdom from John Hopfield in his article titled Now What?', 'start': 4296.972, 'duration': 5.863}, {'end': 4308.822, 'text': 'Choosing problems is the primary determinant of what one accomplishes in science.', 'start': 4304.4, 'duration': 4.422}, {'end': 4313.924, 'text': 'I have generally had a relatively short attention span in science problems.', 'start': 4310.063, 'duration': 3.861}, {'end': 4318.647, 'text': 'Thus, I have always been on the lookout for more interesting questions,', 'start': 4314.465, 'duration': 4.182}, {'end': 4325.71, 'text': 'either as my present ones get worked out or as they get classified by me as intractable, given my particular talents.', 'start': 4318.647, 'duration': 7.063}], 'summary': "Choosing the right problems is crucial in science, as john hopfield emphasizes in his article titled 'now what?'", 'duration': 28.738, 'max_score': 4296.972, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE4296972.jpg'}], 'start': 3698.393, 'title': 'Embracing chaos in neural engineering', 'summary': 'Explores embracing chaos in messy biological systems to improve neural engineering and discusses the search for elegant equations capturing the link between molecules and behavior, the essence of time, and the impact of studying the mind on mortality, also delving into the interconnectedness of biological systems and the human mind while pondering over the meaning of existence.', 'chapters': [{'end': 3782.788, 'start': 3698.393, 'title': 'Neural engineering and chaos', 'summary': 'Discusses the benefits of engineering embracing the large chaos of messy, air-prone biological systems, rather than using highly accurate but complex methods, in order to improve neural engineering.', 'duration': 84.395, 'highlights': ['Understanding how to do things in a more forgiving and neural way would benefit the engineering world by using fewer accurate sensors, such as putting in a pressure sensor or two rather than an array of a gazillion pressure sensors, none of which are accurate.', 'Neurobiology does things differently from an engineer, for example, an engineer would put in six highly accurate stepping motors controlling a limb rather than 100,000 muscle fibers, each of which has to be individually controlled.']}, {'end': 4073.309, 'start': 3783.688, 'title': 'Equations of biology and physics', 'summary': 'Discusses the search for elegant equations to capture the link between molecules and psychological behavior, the essence of time, and the impact of studying the mind on mortality, highlighting the open problem of discovering these equations and the shift towards digital preservation of human facts.', 'duration': 289.621, 'highlights': ['The search for elegant equations to capture the link between molecules and psychological behavior is one of the main open problems of our age. The chapter emphasizes the challenge of discovering elegant equations that can describe the fundamental behavior of molecules and the brain, highlighting it as a key open problem of the current era.', "The impact of studying the mind as a mechanism on mortality is discussed, emphasizing the shift towards digital preservation of human facts. The discussion delves into the realization that what defines individuals is contained in the brain, leading to a heightened awareness of the digital preservation of human facts and the impact of studying the mind on mortality, especially in the context of the body's aging and eventual decay.", "The discussion of elegant equations in physics and the essence of time is linked to the ideas explored in the book 'Einstein's Dreams'. The chapter draws a connection between the exploration of elegant equations in physics and the essence of time, referencing the book 'Einstein's Dreams' and its alternation between chapters on Einstein's life and descriptions of the way time might have been but isn't."]}, {'end': 4359.075, 'start': 4074.35, 'title': 'Exploring the meaning of life and existence', 'summary': 'Delves into the complexities of defining life and exploring the interconnectedness of biological systems and the human mind, while also pondering on the meaning of existence, with insights on the significance of choosing scientific problems for accomplishment.', 'duration': 284.725, 'highlights': ["The chapter discusses the challenges of defining life and the interconnectedness of biological systems, questioning the notion of individuality and the vagueness of the concept of 'species.' It's highlighted that defining life in a living system poses challenges, with differences observed even among organisms of the same nominal species, raising questions about the existence of species in various environments.", 'The transcript explores the interconnectedness of biological systems and the human mind, suggesting a lack of clarity regarding the boundaries of individuality and the essence of thought. The discussion emphasizes the interconnected nature of biological systems and the human mind, blurring the lines between individuality and the essence of thought, raising questions about the boundaries of human existence.', 'Insights are provided on the significance of choosing scientific problems as a primary determinant of scientific accomplishments, emphasizing the pursuit of more interesting questions and the importance of experts in the field. The importance of selecting scientific problems is highlighted as a key determinant of scientific accomplishments, encouraging the pursuit of more intriguing questions and acknowledging the value of experts in experimental and theoretical studies.']}], 'duration': 660.682, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/DKyzcbNr8WE/pics/DKyzcbNr8WE3698393.jpg', 'highlights': ['The pursuit of more interesting questions is highlighted as a key determinant of scientific accomplishments.', 'The interconnected nature of biological systems and the human mind is emphasized, blurring the lines between individuality and the essence of thought.', 'The impact of studying the mind as a mechanism on mortality is discussed, emphasizing the shift towards digital preservation of human facts.', "The discussion of elegant equations in physics and the essence of time is linked to the ideas explored in the book 'Einstein's Dreams'.", 'Understanding how to do things in a more forgiving and neural way would benefit the engineering world by using fewer accurate sensors.', "The chapter discusses the challenges of defining life and the interconnectedness of biological systems, questioning the notion of individuality and the vagueness of the concept of 'species.'"]}], 'highlights': ["John Hopfield's interdisciplinary work in physics, biology, and neuroscience significantly impacted deep learning and machine learning.", 'Receiving the 2019 Franklin Medal in Physics for applying theoretical physics concepts to provide new insights on important biological questions, impacting genetics, neuroscience, and machine learning.', 'The potential of glitches becoming features in biological neural networks due to evolutionary processes.', 'The essential role of feedback systems in computation, despite complexity and exponential expansion.', 'The massive scale of physical systems like laptops with approximately 10^10 transistors', 'The interconnected nature of biological systems and the human mind is emphasized, blurring the lines between individuality and the essence of thought.']}