On-line learning dynamics of ReLU neural networks using statistical physics techniques

Abstract

We introduce exact macroscopic on-line learning dynamics of two-layer neural networks with ReLU units in the form of a system of differential equations, using techniques borrowed from statistical physics. For the first experiments, numerical solutions reveal similar behavior compared to sigmoidal activation researched in earlier work. In these experiments the theoretical results show good correspondence with simulations. In ove-rrealizable and unrealizable learning scenarios, the learning behavior of ReLU networks shows distinctive characteristics compared to sigmoidal networks.

Publication
Proc. European Symposium on Artificial Neural Networks (ESANN) 2019, Bruges/Belgium
Avatar
Michiel Straat
PhD Student Intelligent Systems

My research interests include Machine Learning, Computational Intelligence and Statistical Physics of Learning.