English

A Hybrid Stochastic Optimization Framework for Stochastic Composite Nonconvex Optimization

Optimization and Control 2020-05-05 v2 Machine Learning Machine Learning

Abstract

We introduce a new approach to develop stochastic optimization algorithms for a class of stochastic composite and possibly nonconvex optimization problems. The main idea is to combine two stochastic estimators to create a new hybrid one. We first introduce our hybrid estimator and then investigate its fundamental properties to form a foundational theory for algorithmic development. Next, we apply our theory to develop several variants of stochastic gradient methods to solve both expectation and finite-sum composite optimization problems. Our first algorithm can be viewed as a variant of proximal stochastic gradient methods with a single-loop, but can achieve O(σ3ε1+σε3)\mathcal{O}(\sigma^3\varepsilon^{-1} + \sigma \varepsilon^{-3})-oracle complexity bound, matching the best-known ones from state-of-the-art double-loop algorithms in the literature, where σ>0\sigma > 0 is the variance and ε\varepsilon is a desired accuracy. Then, we consider two different variants of our method: adaptive step-size and restarting schemes that have similar theoretical guarantees as in our first algorithm. We also study two mini-batch variants of the proposed methods. In all cases, we achieve the best-known complexity bounds under standard assumptions. We test our methods on several numerical examples with real datasets and compare them with state-of-the-arts. Our numerical experiments show that the new methods are comparable and, in many cases, outperform their competitors.

Keywords

Cite

@article{arxiv.1907.03793,
  title  = {A Hybrid Stochastic Optimization Framework for Stochastic Composite Nonconvex Optimization},
  author = {Quoc Tran-Dinh and Nhan H. Pham and Dzung T. Phan and Lam M. Nguyen},
  journal= {arXiv preprint arXiv:1907.03793},
  year   = {2020}
}

Comments

49 pages, 2 tables, 9 figures

R2 v1 2026-06-23T10:15:15.534Z