Random matrix analysis of deep neural network weight matrices
This work provides theoretical insights into how neural networks store information, which is incremental as it applies existing random matrix methods to a new context in deep learning.
The authors analyzed weight matrices of trained deep neural networks using random matrix theory, finding that most singular values and eigenvectors follow universal predictions, indicating randomness, while only the largest singular values deviate, potentially encoding learned information, and this approach distinguished between lazy and rich learning regimes.
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.