Saved in:
Bibliographic Details
Main Authors: Glasgow, Margalit, Wu, Denny, Bruna, Joan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.13110
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908556215713792
author Glasgow, Margalit
Wu, Denny
Bruna, Joan
author_facet Glasgow, Margalit
Wu, Denny
Bruna, Joan
contents We study the approximation gap between the dynamics of a polynomial-width neural network and its infinite-width counterpart, both trained using projected gradient descent in the mean-field scaling regime. We demonstrate how to tightly bound this approximation gap through a differential equation governed by the mean-field dynamics. A key factor influencing the growth of this ODE is the local Hessian of each particle, defined as the derivative of the particle's velocity in the mean-field dynamics with respect to its position. We apply our results to the canonical feature learning problem of estimating a well-specified single-index model; we permit the information exponent to be arbitrarily large, leading to convergence times that grow polynomially in the ambient dimension $d$. We show that, due to a certain ``self-concordance'' property in these problems -- where the local Hessian of a particle is bounded by a constant times the particle's velocity -- polynomially many neurons are sufficient to closely approximate the mean-field dynamics throughout training.
format Preprint
id arxiv_https___arxiv_org_abs_2504_13110
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Propagation of Chaos in One-hidden-layer Neural Networks beyond Logarithmic Time
Glasgow, Margalit
Wu, Denny
Bruna, Joan
Machine Learning
We study the approximation gap between the dynamics of a polynomial-width neural network and its infinite-width counterpart, both trained using projected gradient descent in the mean-field scaling regime. We demonstrate how to tightly bound this approximation gap through a differential equation governed by the mean-field dynamics. A key factor influencing the growth of this ODE is the local Hessian of each particle, defined as the derivative of the particle's velocity in the mean-field dynamics with respect to its position. We apply our results to the canonical feature learning problem of estimating a well-specified single-index model; we permit the information exponent to be arbitrarily large, leading to convergence times that grow polynomially in the ambient dimension $d$. We show that, due to a certain ``self-concordance'' property in these problems -- where the local Hessian of a particle is bounded by a constant times the particle's velocity -- polynomially many neurons are sufficient to closely approximate the mean-field dynamics throughout training.
title Propagation of Chaos in One-hidden-layer Neural Networks beyond Logarithmic Time
topic Machine Learning
url https://arxiv.org/abs/2504.13110