Saved in:
Bibliographic Details
Main Authors: Olsen, Brian Richard, Fatehmanesh, Sam, Xiao, Frank, Kumarappan, Adarsh, Gajula, Anirudh
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.12709
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917256009613312
author Olsen, Brian Richard
Fatehmanesh, Sam
Xiao, Frank
Kumarappan, Adarsh
Gajula, Anirudh
author_facet Olsen, Brian Richard
Fatehmanesh, Sam
Xiao, Frank
Kumarappan, Adarsh
Gajula, Anirudh
contents Deep neural networks have revolutionized machine learning, yet their training dynamics remain theoretically unclear-we develop a continuous-time, matrix-valued stochastic differential equation (SDE) framework that rigorously connects the microscopic dynamics of SGD to the macroscopic evolution of singular-value spectra in weight matrices. We derive exact SDEs showing that squared singular values follow Dyson Brownian motion with eigenvalue repulsion, and characterize stationary distributions as gamma-type densities with power-law tails, providing the first theoretical explanation for the empirically observed 'bulk+tail' spectral structure in trained networks. Through controlled experiments on transformer and MLP architectures, we validate our theoretical predictions and demonstrate quantitative agreement between SDE-based forecasts and observed spectral evolution, providing a rigorous foundation for understanding why deep learning works.
format Preprint
id arxiv_https___arxiv_org_abs_2507_12709
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From SGD to Spectra: A Theory of Neural Network Weight Dynamics
Olsen, Brian Richard
Fatehmanesh, Sam
Xiao, Frank
Kumarappan, Adarsh
Gajula, Anirudh
Machine Learning
Deep neural networks have revolutionized machine learning, yet their training dynamics remain theoretically unclear-we develop a continuous-time, matrix-valued stochastic differential equation (SDE) framework that rigorously connects the microscopic dynamics of SGD to the macroscopic evolution of singular-value spectra in weight matrices. We derive exact SDEs showing that squared singular values follow Dyson Brownian motion with eigenvalue repulsion, and characterize stationary distributions as gamma-type densities with power-law tails, providing the first theoretical explanation for the empirically observed 'bulk+tail' spectral structure in trained networks. Through controlled experiments on transformer and MLP architectures, we validate our theoretical predictions and demonstrate quantitative agreement between SDE-based forecasts and observed spectral evolution, providing a rigorous foundation for understanding why deep learning works.
title From SGD to Spectra: A Theory of Neural Network Weight Dynamics
topic Machine Learning
url https://arxiv.org/abs/2507.12709