A mean-field limit for certain deep neural networks

MaRDI QIDQ6319780zbMATH OpenFDO

Authors Dyego Araújo, Roberto I. Oliveira, Daniel Yukimura

Publication date 1 June 2019

Mathematics Subject Classification ID

Interacting random processes; statistical mechanics type models; percolation theory (60K35) Neural networks for/in biological studies, artificial life and related topics (92B20)

Abstract: Understanding deep neural networks (DNNs) is a key challenge in the theory of machine learning, with potential applications to the many fields where DNNs have been successfully used. This article presents a scaling limit for a DNN being trained by stochastic gradient descent. Our networks have a fixed (but arbitrary) number $Lgeq2$ of inner layers; $Ngg1$ neurons per layer; full connections between layers; and fixed weights (or "random features" that are not trained) near the input and output. Our results describe the evolution of the DNN during training in the limit when $No+infty$, which we relate to a mean field model of McKean-Vlasov type. Specifically, we show that network weights are approximated by certain "ideal particles" whose distribution and dependencies are described by the mean-field model. A key part of the proof is to show existence and uniqueness for our McKean-Vlasov problem, which does not seem to be amenable to existing theory. Our paper extends previous work on the $L=1$ case by Mei, Montanari and Nguyen; Rotskoff and Vanden-Eijnden; and Sirignano and Spiliopoulos. We also complement recent independent work on $L>1$ by Sirignano and Spiliopoulos (who consider a less natural scaling limit) and Nguyen (who nonrigorously derives similar results).

This page was built for publication: A mean-field limit for certain deep neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6319780)