Random matrix analysis of deep neural network weight matrices

TL;DR

This paper applies random matrix theory to analyze trained deep neural network weight matrices, revealing that most singular values behave randomly, while the largest singular values may encode learned information, and distinguishes different training regimes.

Contribution

It demonstrates that the spectral statistics of neural network weights largely follow universal RMT predictions, with deviations indicating learned features, and differentiates training regimes based on spectral analysis.

Findings

Most singular values follow RMT predictions

Eigenvectors of largest singular values deviate from randomness

Spectral distribution distinguishes training regimes

Abstract

Neural networks have been used successfully in a variety of fields, which has led to a great deal of interest in developing a theoretical understanding of how they store the information needed to perform a particular task. We study the weight matrices of trained deep neural networks using methods from random matrix theory (RMT) and show that the statistics of most of the singular values follow universal RMT predictions. This suggests that they are random and do not contain system specific information, which we investigate further by comparing the statistics of eigenvector entries to the universal Porter-Thomas distribution. We find that for most eigenvectors the hypothesis of randomness cannot be rejected, and that only eigenvectors belonging to the largest singular values deviate from the RMT prediction, indicating that they may encode learned information. In addition, a comparison with RMT predictions also allows to distinguish networks trained in different learning regimes - from lazy to rich learning. We analyze the spectral distribution of the large singular values using the Hill estimator and find that the distribution cannot in general be characterized by a tail index, i.e. is not of power law type.