English

Bandwidth-based Step-Sizes for Non-Convex Stochastic Optimization

Machine Learning 2021-10-13 v2 Optimization and Control

Abstract

Many popular learning-rate schedules for deep neural networks combine a decaying trend with local perturbations that attempt to escape saddle points and bad local minima. We derive convergence guarantees for bandwidth-based step-sizes, a general class of learning rates that are allowed to vary in a banded region. This framework includes many popular cyclic and non-monotonic step-sizes for which no theoretical guarantees were previously known. We provide worst-case guarantees for SGD on smooth non-convex problems under several bandwidth-based step sizes, including stagewise 1/t1/\sqrt{t} and the popular step-decay (constant and then drop by a constant), which is also shown to be optimal. Moreover, we show that its momentum variant converges as fast as SGD with the bandwidth-based step-decay step-size. Finally, we propose novel step-size schemes in the bandwidth-based family and verify their efficiency on several deep neural network training tasks.

Keywords

Cite

@article{arxiv.2106.02888,
  title  = {Bandwidth-based Step-Sizes for Non-Convex Stochastic Optimization},
  author = {Xiaoyu Wang and Mikael Johansson},
  journal= {arXiv preprint arXiv:2106.02888},
  year   = {2021}
}
R2 v1 2026-06-24T02:52:04.450Z