Evaluating Singular Value Thresholds for DNN Weight Matrices based on Random Matrix Theory
This work addresses the challenge of efficient model compression for deep learning practitioners, but it is incremental as it builds on existing random matrix theory approaches.
The study tackled the problem of determining thresholds for removing singular values in low-rank approximations of deep neural network weight matrices by modeling them as signal plus noise, and it proposed an evaluation metric based on cosine similarity to compare two threshold estimation methods in numerical experiments.
This study evaluates thresholds for removing singular values from singular value decomposition-based low-rank approximations of deep neural network weight matrices. Each weight matrix is modeled as the sum of signal and noise matrices. The low-rank approximation is obtained by removing noise-related singular values using a threshold based on random matrix theory. To assess the adequacy of this threshold, we propose an evaluation metric based on the cosine similarity between the singular vectors of the signal and original weight matrices. The proposed metric is used in numerical experiments to compare two threshold estimation methods.