What is Deep in Deep Learning?
Wednesday March 22, 2023 - 14.30 (GMT+1)
Boris Hanin
Princeton University
Embedded Files
Depth is key to the success of modern neural networks. Mathematically, however, it is not obvious why and when deeper is better. This talk will survey several lines of work, spanning approximation theory, probability, statistical physics, and optimization, that have sought to answer such questions.
Page updated
Google Sites
Report abuse