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

Straat, M. & Biehl, M., 18-Mar-2019, In : ArXiv.

Research output: Contribution to journalArticleAcademicpeer-review

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.
Original languageEnglish
Publication statusPublished - 18-Mar-2019


  • cs.LG, cond-mat.dis-nn, stat.ML
Related Activities
  1. European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

    Michiel Straat (Speaker), Michael Biehl (Participant)

    Activity: Participating in or organising an eventParticipation in conferenceAcademic

View all (1) »

Download statistics

No data available

ID: 78117626