On the Convergence and Complexity of Proximal Gradient and Accelerated Proximal Gradient Methods under Adaptive Gradient Estimation

TL;DR

This paper introduces adaptive gradient estimation techniques within proximal gradient methods for composite optimization, achieving optimal convergence rates even with biased or stochastic gradient estimates across various convexity settings.

Contribution

It develops new proximal algorithms that adaptively refine gradient estimates, ensuring optimal complexity and convergence for nonconvex, convex, and strongly convex problems.

Findings

Achieves optimal iteration complexity for first-order methods.

Demonstrates efficiency with biased and unbiased gradient estimates.

Validates theoretical results through numerical experiments.

Abstract

In this paper, we propose a proximal gradient method and an accelerated proximal gradient method for solving composite optimization problems, where the objective function is the sum of a smooth and a convex, possibly nonsmooth, function. We consider settings where the smooth component is either a finite-sum function or an expectation of a stochastic function, making it computationally expensive or impractical to evaluate its gradient. To address this, we utilize gradient estimates within the proximal gradient framework. Our methods dynamically adjust the accuracy of these estimates, increasing it as the iterates approach a solution, thereby enabling high-precision solutions with minimal computational cost. We analyze the methods when the smooth component is nonconvex, convex, or strongly convex, using a biased gradient estimate. In all cases, the methods achieve the optimal iteration complexity for first-order methods. When the gradient estimate is unbiased, we further refine the analysis to show that the methods simultaneously achieve optimal iteration complexity and optimal complexity in terms of the number of stochastic gradient evaluations. Finally, we validate our theoretical results through numerical experiments.