Faster Gradient-Free Proximal Stochastic Methods for Nonconvex Nonsmooth Optimization

TL;DR

This paper introduces faster zeroth-order proximal stochastic algorithms with variance reduction techniques for nonconvex nonsmooth optimization, achieving improved convergence rates over previous methods.

Contribution

The paper develops ZO-ProxSVRG and ZO-ProxSAGA algorithms with $O(1/T)$ convergence rates, addressing the unbiased gradient estimation challenge in zeroth-order methods.

Findings

Algorithms outperform existing zeroth-order methods in convergence speed.

Theoretical proof of $O(1/T)$ convergence rate.

Experimental results confirm faster convergence.

Abstract

Proximal gradient method has been playing an important role to solve many machine learning tasks, especially for the nonsmooth problems. However, in some machine learning problems such as the bandit model and the black-box learning problem, proximal gradient method could fail because the explicit gradients of these problems are difficult or infeasible to obtain. The gradient-free (zeroth-order) method can address these problems because only the objective function values are required in the optimization. Recently, the first zeroth-order proximal stochastic algorithm was proposed to solve the nonconvex nonsmooth problems. However, its convergence rate is $O(\frac{1}{\sqrt{T}})$ for the nonconvex problems, which is significantly slower than the best convergence rate $O(\frac{1}{T})$ of the zeroth-order stochastic algorithm, where $T$ is the iteration number. To fill this gap, in the paper, we propose a class of faster zeroth-order proximal stochastic methods with the variance reduction techniques of SVRG and SAGA, which are denoted as ZO-ProxSVRG and ZO-ProxSAGA, respectively. In theoretical analysis, we address the main challenge that an unbiased estimate of the true gradient does not hold in the zeroth-order case, which was required in previous theoretical analysis of both SVRG and SAGA. Moreover, we prove that both ZO-ProxSVRG and ZO-ProxSAGA algorithms have $O(\frac{1}{T})$ convergence rates. Finally, the experimental results verify that our algorithms have a faster convergence rate than the existing zeroth-order proximal stochastic algorithm.