English

Probabilistic Guarantees of Stochastic Recursive Gradient in Non-Convex Finite Sum Problems

Machine Learning 2024-01-31 v1 Machine Learning Optimization and Control Statistics Theory Statistics Theory

Abstract

This paper develops a new dimension-free Azuma-Hoeffding type bound on summation norm of a martingale difference sequence with random individual bounds. With this novel result, we provide high-probability bounds for the gradient norm estimator in the proposed algorithm Prob-SARAH, which is a modified version of the StochAstic Recursive grAdient algoritHm (SARAH), a state-of-art variance reduced algorithm that achieves optimal computational complexity in expectation for the finite sum problem. The in-probability complexity by Prob-SARAH matches the best in-expectation result up to logarithmic factors. Empirical experiments demonstrate the superior probabilistic performance of Prob-SARAH on real datasets compared to other popular algorithms.

Keywords

Cite

@article{arxiv.2401.15890,
  title  = {Probabilistic Guarantees of Stochastic Recursive Gradient in Non-Convex Finite Sum Problems},
  author = {Yanjie Zhong and Jiaqi Li and Soumendra Lahiri},
  journal= {arXiv preprint arXiv:2401.15890},
  year   = {2024}
}

Comments

41 pages, 3 figures, accepted to PAKDD 2024