Giovedí
20 febbraio 2020
Ore 15:00
Aula "A. Rostagni"
Prof. Riccardo Zecchina
Theoretical Physics, Chair in Machine Learning, Bocconi University,
Milano, Italy
Deep neural
networks (DNN) are becoming fundamental learning devices
for extracting information from data in a variety of real-world
applications and in natural and social sciences. The learning
process in DNN consists of finding a minimizer of a highly
non-convex loss function that measures how well the data
are classified. This optimization task is typically solved
by tuning millions of parameters by stochastic gradient
algorithms.
The learning process is often observed to be able to find
good minimizers without getting stuck in local critical
points, and that such minimizers are often satisfactory
at avoiding overfitting. How these two features can be
kept under control in nonlinear devices composed of millions
of tunable connections is a profound and far reaching
open question.
Here we
discuss how to use the out-of-equilibrium techniques of
statistical physics to study the peculiar geometrical
structure of neural networks and to design novel
learning algorithms.
