English

Large Learning Rate Tames Homogeneity: Convergence and Balancing Effect

Machine Learning 2022-03-01 v2 Dynamical Systems Optimization and Control

Abstract

Recent empirical advances show that training deep models with large learning rate often improves generalization performance. However, theoretical justifications on the benefits of large learning rate are highly limited, due to challenges in analysis. In this paper, we consider using Gradient Descent (GD) with a large learning rate on a homogeneous matrix factorization problem, i.e., minX,YAXYF2\min_{X, Y} \|A - XY^\top\|_{\sf F}^2. We prove a convergence theory for constant large learning rates well beyond 2/L2/L, where LL is the largest eigenvalue of Hessian at the initialization. Moreover, we rigorously establish an implicit bias of GD induced by such a large learning rate, termed 'balancing', meaning that magnitudes of XX and YY at the limit of GD iterations will be close even if their initialization is significantly unbalanced. Numerical experiments are provided to support our theory.

Keywords

Cite

@article{arxiv.2110.03677,
  title  = {Large Learning Rate Tames Homogeneity: Convergence and Balancing Effect},
  author = {Yuqing Wang and Minshuo Chen and Tuo Zhao and Molei Tao},
  journal= {arXiv preprint arXiv:2110.03677},
  year   = {2022}
}

Comments

45 pages

R2 v1 2026-06-24T06:43:00.939Z