Shengyu Cai

Federated Spectral Regularization for Convergence Acceleration: A Random Matrix Theory Perspective

This paper analyzes the convergence problem from a spectral perspective, revealing that non-IID data causes spectral diffusion in the gradient covariance matrix and degrades convergence, and proposes federated spectral regularization (Fed-SR), a computationally efficient method that indirectly constrains spectral sprea...