W1S2 u2013 Yann LeCun.pdf - Big Ideas Course Series...

This preview shows page 1 - 11 out of 39 pages.

Artificial Intelligence Yann LeCun NYU - Courant Institute & Center for Data Science Big Ideas: AI, Spring 2021 Big Ideas Course Series
What is AI? And what is it used for?
Y. LeCun Model of The World configuration Critic Cost Actor Perception sensors actuators What is Artificial Intelligence? Getting machines to perform tasks normally attributed to animals and humans Perception, representation, memory, reasoning/predicting, action. Perception : vision, audition, touch, radar, sonar, lidar, other sensors... Representation : Objects in a scene, meaning of a text Memory : storing and retrieving representations Reasoning : manipulating representations, predicting, planning a sequence of actions, search. Action : producing an answer, moving actuators
Y. LeCun Applications of AI: Protecting People & the Environment Transportation Driving assistance / autonomous driving Medicine Medical imaging Diagnostic aid, patient care Drug discovery On-line Safety / Security Filtering harmful/hateful content Filtering dangerous misinformation Environmental monitoring AI saves lives
Y. LeCun Applications of AI: Connecting people & Information Connecting people with knowledge “Wealth of information, scarcity of attention” [Raj Reddy] Content filtering, ranking, indexing, search, translation Augmented reality Connecting people with each other Language translation, content filtering and ranking, Virtual reality and telepresence Helping the economy Industrial process optimization, Manufacturing, logistics Connecting vendors and customers
Y. LeCun Deep Learning Connects People to knowledge & to each other Facebook, Google, YouTube, Amazon,...are built around Deep learning Take Deep Learning out of them, and they crumble. Dealing with the information deluge Search, retrieval, ranking, question-answering Requires machines to understand content Translation / transcription / accessibility language ↔ language, text ↔ speech, image → text 3 billion people can’t use technology today. 800 million are illiterate 300 million are visually impaired People speak different languages
Y. LeCun Applications of AI: Science & Medicine Understanding intelligence “I can only understand what I can build” (paraphrasing Richard Feynman) How do humans and animal learn? How could machine learn like humans and animals? AI for science Neuroscience Biology, genomics Physics Chemistry, Material science Social sciences [Eickenberg 2016] [Komiske 2016] [He 2019]
A bit of History Two major currents in AI: 1. based on logic and symbol manipulation. 2. based on learning, inspired by neuroscience.
Y. LeCun AI timeline: Cybernetics , Neuroscience , AI/Logic/Algorithms 1943: McCulloch & Pitts, binary neurons can do logic 1947: Donald Hebb, Hebbian synaptic plasticity 1948: Norbert Wiener, cybernetics 1957: Frank Rosenblatt, Perceptron, learning 1958: John McCarthy, Lisp language 1959: Simon & Newell, General Problem Solver 1959: Arthur Samuel, checker player, tree search 1961: Bernie Widrow, Adaline Wiener Rosenblatt Newell & Simon
Y. LeCun AI timeline: neural nets , neuroscience , logic-based AI 1962: Hubel & Wiesel, visual cortex architecture 1969: Minsky & Papert, limits of the Perceptron

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture