Results
83 articlesSort
1:17:29EN
Lex Fridman· Sep 23, 2019Podcast · 2 speakersThe Algorithm Is Already Better. The Clinic Still Uses 1967.Guest Regina Barzilay · Host Lex FridmanMIT's Regina Barzilay on why deep learning already beats the breast-cancer 'density' score — and why data, liability, and incentives still will not move.
1:59:49ENLex Fridman· Sep 14, 2019Podcast · 2 speakersIntelligence is not the height of a building: François Chollet on Keras, deep learning, and why AGI will not explodeGuest François Chollet · Host Lex FridmanThe Keras creator tells Lex Fridman that intelligence lives in a brain, a body, and a problem, that science already is a recursively self-improving superhuman system with linear output, and that the real near-term danger is the newsfeed.
1:44:11ENLex Fridman· Aug 27, 2019Podcast · 2 speakersA Single GPU Is Enough: Jeremy Howard on Teaching Deep LearningGuest Jeremy Howard · Host Lex FridmanThe fast.ai founder tells Lex Fridman why domain experts, not more papers or more TPUs, will decide whether deep learning actually helps people.
1:10:58ENLex Fridman· Jun 3, 2019Podcast · 2 speakersThe Hadoop Lesson That Made Google Give TensorFlow AwayGuest Rajat Monga · Host Lex FridmanIn November 2015, Google put its core internal machine-learning system on GitHub. Rajat Monga explains why a company of that size would hand over the hard-to-copy part.
1:46:01ENLex Fridman· Apr 29, 2019Podcast · 2 speakersThe Night AlphaStar Beat TLO, the Team Went to a BarGuest Oriol Vinyals · Host Lex FridmanOriol Vinyals on how DeepMind turned StarCraft II into a sequence problem, built a league of agents, and still refuses to call the game solved.
1:08:37ENLex Fridman· Apr 18, 2019Podcast · 2 speakersA Bar in Montreal, Then a Night of Code: Ian Goodfellow on Why GANs WorkGuest Ian Goodfellow · Host Lex FridmanLex Fridman talks with Ian Goodfellow about the 2014 invention of generative adversarial networks, adversarial examples as a security problem, and what still has to be built before machines can hold a fact the way a person does.
1:20:21ENLex Fridman· Jan 19, 2019Podcast · 2 speakersEinstein Was the Worst of Five Ph.D. StudentsGuest Tomaso Poggio · Host Lex FridmanMIT neuroscientist Tomaso Poggio on why intelligence is science's deepest problem, and what brains still have to teach machines.
46:25ENLex Fridman· Jan 17, 2019PodcastDeep Learning at the Edge of 2019: BERT, Autopilot, and the Ideas That Broke ThroughHost Lex FridmanA tour of the architectural, applied, and automation breakthroughs that defined deep learning in 2018 and set up 2019.
1:19:59ENLex Fridman· Dec 23, 2018Podcast · 2 speakersSchmidhuber’s bet: true meta-learning rewrites the learner, not the last layerGuest Juergen Schmidhuber · Host Lex FridmanJürgen Schmidhuber tells Lex Fridman how Gödel machines, compression-as-science, Power Play, LSTMs, and world models all descend from a teenage plan to multiply one person’s curiosity into a better physicist than he could ever be.
42:45ENLex Fridman· Dec 16, 2018Podcast · 2 speakersA one-legged robot learned a backflip from "this is better"Guest Pieter Abbeel · Host Lex FridmanPieter Abbeel tells Lex Fridman why preference comparisons, self-play, ten-minute demos, and messy ensembles of simulators are the practical path through deep reinforcement learning.
54:03ENLex Fridman· Nov 16, 2018Podcast · 2 speakersThe Hardest Part of Learning Was Left Out of the TextbooksGuest Vladimir Vapnik · Host Lex FridmanVladimir Vapnik tells Lex Fridman why intelligence is the work of inventing predicates that shrink the admissible set of functions, not of fitting a rule to a mountain of data.
42:18ENLex Fridman· Oct 20, 2018Podcast · 2 speakersAn Adult Can Change Last Year's Mind. A Network Still Cannot.Guest Yoshua Bengio · Host Lex FridmanYoshua Bengio tells Lex Fridman that the missing pieces of deep learning sit in credit assignment, training objectives, and a kind of compositionality borrowed from classical AI — not in ten thousand layers.
1:29:03ENLex Fridman· Sep 27, 2016PodcastFrom Word Vectors to Memory Networks: The Building Blocks of Deep NLPHost Lex FridmanRichard Socher explains how word embeddings, recurrent models, GRUs, pointer mixtures, and dynamic memory networks turn language tasks into trainable end-to-end systems.
1:00:51ENLex FridmanPodcastDeep Learning From the Ground Up: Hugo Larochelle's One-Hour Tour of Neural Network FoundationsHost Lex FridmanA fast, equation-level walk through feed-forward networks, backpropagation, practical training tricks, dropout and batch normalization, from the 2016 Deep Learning School.
1:07:29ENLex FridmanPodcastDeep Reinforcement Learning from First Principles: Rewards, Q-Learning, Policy Gradients and AlphaZeroHost Lex FridmanLex Fridman's MIT 6.S091 opening lecture walks from the basic RL framework through DQN, actor-critic methods and AlphaZero to the unsolved problem of moving from simulation to the real world.
9:59ENLex FridmanPodcastIs Language or Vision Harder? Ilya Sutskever on the Unity of Machine LearningHost Lex FridmanIn a short clip from his 2020 conversation with Lex Fridman, Ilya Sutskever explains why vision, language and reinforcement learning share the same few principles, and why asking which problem is harder is the wrong question.
15:01ENLex FridmanPodcast · 2 speakersDeep Learning Is a System 1 Machine: Daniel Kahneman on What AI Still LacksGuest Daniel Kahneman · Host Lex FridmanThe Nobel-winning psychologist tells Lex Fridman how his fast-and-slow thinking framework maps onto deep learning, and why driving among pedestrians may require understanding people.
1:25:16ENLex FridmanPodcastFrom Hand-Built Features to ResNets: Andrej Karpathy's Crash Course in Convolutional NetworksHost Lex FridmanA 2016 Deep Learning School talk covering how convnets work, how the ImageNet winners evolved, and how to apply them without being a hero.
54:33ENLex FridmanPodcast · 2 speakersVision or Lidar, Human or Machine: Lex Fridman's 2019 Map of the Self-Driving RaceGuest Self-Driving Cars · Host Lex FridmanThe opening lecture of MIT's 6.S094 course surveys 2018's milestones, the industry's optimistic predictions, and the beliefs about driving and humans that will decide who wins.
1:24:43ENLex FridmanPodcastFrom Sparse Coding to GANs: A Tour of Unsupervised Deep Learning in 2016Host Lex FridmanRuslan Salakhutdinov walks through autoencoders, Boltzmann machines, variational autoencoders and adversarial networks at the Deep Learning School, with candid notes on what did and did not work.
58:46ENLex FridmanPodcastCaging the Neural Network: How Aptiv Thinks About Safe Self-DrivingHost Lex FridmanAptiv Autonomous Mobility's Karl Iagnemma and Oscar Beijbom explain why end-to-end learning is hard to trust in cars, and show how PointPillars and nuScenes push lidar perception forward.
1:15:59ENLex FridmanPodcastSteering Through Time: How Recurrent Networks and LSTMs Learn from SequencesHost Lex FridmanLex Fridman's MIT 6.S094 lecture walks from backpropagation basics to LSTMs and the way they are used to predict steering in self-driving cars.
18:46ENLex FridmanPodcastDeep Learning Plus One Idea: Ilya Sutskever on Building AGIHost Lex FridmanIn a May 2020 conversation, Sutskever discusses self-play, simulation, bodies and consciousness, tests of intelligence, and how an AGI might be governed and aligned.
1:31:52ENLex FridmanPodcastFrom Phonemes to CTC: How Deep Learning Rebuilt Speech RecognitionHost Lex FridmanBaidu's Adam Coates walks through the traditional speech pipeline, the CTC-based end-to-end engine that replaced much of it, and what it takes to scale and deploy it.Load moreShowing 24 of 83