Artificial intelligence: success, limits, myths and threats (Lecture 1) by Marc Mézard

INFOSYS-ICTS TURING LECTURES ARTIFICIAL INTELLIGENCE: SUCCESS, LIMITS, MYTHS AND THREATS SPEAKER: Marc Mézard (Director of Ecole normale supérieure - PSL University ) DATE: 06 January 2020, 16:00 to 17:30 VENUE: Chandrasekhar Auditorium, ICTS-TIFR, Bengaluru Lecture 1 (Public Lecture): 6 January 2020, 4:00 PM Title: Artificial intelligence: success, limits, myths and threats Abstract: Artificial Intelligence is about to have a dramatic impact on many sectors of human activity. In the last ten years, thanks to the development of machine learning in “deep networks”, we have experienced spectacular breakthroughs in diverse applications such as automatic interpretation of images, speech recognition, consumer profiling, or go and chess playing. Algorithms are now competing with the best professionals at analyzing skin cancer symptoms or detecting specific anomalies in radiology; and much more is to come. Worrisome perspectives are frequently raised, from massive job destruction to autonomous decision-making “warrior” robots. In this talk, we shall open the black box of deep networks and explore how they are programmed to learn from data by themselves. This will allow us to understand their limits, to question whether their achievements have anything to do with “intelligence”, and to reflect on the foundations of scientific intelligence. Venue: Chandrasekhar Auditorium, ICTS-TIFR, Bengaluru Lecture 2: Tuesday 7th January 2020, 4:00 PM Title: Spin glasses and hard optimization problems Venue: Madhava Hall, ICTS-TIFR, Bengaluru Lecture 3: Wednesday 8th January 2020, 4:00 PM Title: Statistical physics and statistical inference Venue: Madhava Hall, ICTS-TIFR, Bengaluru This lecture is part of the 'Statistical Physics of Machine Learning' discussion meeting. Table of Contents (powered by https://videoken.com) 0:00:00 Artificial intelligence: success, limits, myths and threats (Lecture 1) 0:01:04 ICTS 0:01:34 ICTS Campus in Bangalore 0:01:42 What is the Goal of the ICTS? 0:02:41 Enabled by 3 interactive missions: 0:03:12 During the past decade ICTS has achieved some measure of success in all its three missions! 0:04:27 Programs: 0:07:47 Sample Programs... 0:08:48 Programs in Machine Learning 0:11:45 ICTS-Infosys (Foundation) Lecture series: 0:12:19 Research 0:13:24 ICTS People: Faculty 0:13:49 Students and Postdocs 0:14:29 ICTS as a platform for new initiatives: 0:15:31 Science Outreach 0:16:42 Public Lectures 0:16:58 Kaapi with Kuriosity 0:17:54 Abdus Salam Memorial Lectures 0:18:06 Einstein Lectures 0:18:39 Vishveshwara Lectures 0:19:36 D.D. Kosambi Lectures 0:20:27 Mathematics of Planet Earth (MPE) 2013, Bengaluru 0:20:33 Bangalore Area Science Habba 0:20:46 Mathematics Circles 0:21:16 ICTS Organization 0:21:51 ICTS Resources 0:22:58 Thank You! 0:26:55 Artificial Intelligence: Success, Limits, Myths and Threats 0:27:09 Chapter One - Myths and Reality 0:29:40 The new era of AI 0:31:02 ImageNet Database and Challenge 0:34:20 Convoy of self-driving trucks completes first European cross-border trip 0:38:14 The new era of AI 0:39:20 2 - Language understanding 0:40:51 AlphaGo 0:42:04 July 2019 : Pluribus 0:42:55 Chapter Two - Machine learning 0:43:36 Machine Learning 0:45:19 Test phase=present new picture, that the machine has not yet seen 0:45:49 Chapter Three - The Machines: Artificial neural networks 0:45:57 Everyone recognizes 0:47:01 Artificial neural networks 0:49:32 Frank Rosenblatt's perception 0:50:27 What is new since Rosenblatt's perceptron? 0:51:19 Neural network reading digits 0:53:12 Performance on handwritten digits 0:54:33 Deep neural networks 0:55:13 Bigger networks, more parameters. Larger database! 0:55:19 New computing paradigm. Collective representation of information, going to larger scales. Robust. 0:56:35 Chapter Four - Why deep networks are not (yet?) a panacea 0:57:02 Three main problems: 0:57:23 1- Huge amount of labelled data is necessary for learning in deep networks 0:59:34 Oh, look at ko bamoule! Do you see ko bamoule? 1:11:00 Chapter Five - About (scientific) Intelligence 1:11:26 Quote from Chris Anderson -The end of Theory: The data deluge makes the scientific method obsolete 1:12:55 Thought experiment : 1:16:01 We are still very very far from General Artificial Intelligence 1:16:30 Conclusion - So, what is going to happen? 1:16:49 Predicting (the future) 1:19:15 Predicting the future (?) 1:21:28 A major concern for the present: 1:22:46 In 2018: 1:24:47 Take-home message 1:25:38 The End 1:25:51 Q&A

Видео Artificial intelligence: success, limits, myths and threats (Lecture 1) by Marc Mézard автора Голливудские шоу-фестивали