Physics and machine learning

DEC 15, 2017 | 9:30 AM TO 6:00 PM



The Graduate Center
365 Fifth Avenue


4102: Science Center


December 15, 2017: 9:30 AM-6:00 PM




Initiative for the Theoretical Sciences and CUNY doctoral programs in Physics and Biology


Recent years have witnessed a revolution in statistical inference and machine learning, in nearly every area of the subject from visual object recognition and natural language processing to reinforcement learning. While many of these advances involve deep connections to statistical physics, a general theoretical framework for understanding why these techniques work and how to improve them has yet to emerge. It is therefore natural to ask what further insights remain to be found at the intersection of machine learning and fields such as statistical physics, condensed matter, and quantum information. In parallel, researchers are using machine learning techniques to gain insight into physical systems. This symposium brings together four researchers working at the interface between machine learning and various areas of physics.

Statistical physics of learning a rule: Decades old story continued
Lenka Zdeborova, CNRS, Saclay

Expressiveness of Convolutional Networks via Quantum Entanglement
Nadav Cohen, Hebrew U. and IAS

Learning to navigate turbulent environments
Massimo Vergassola, UCSD

Reinforcement Learning, Optimal Control, and Physics
Pankaj Mehta, Boston U.

View/download the workshop schedule