Online Stochastic Gradient Methods Under Sub-Weibull Noise and the Polyak-{\L}ojasiewicz Condition

TL;DR

This paper analyzes the convergence of online stochastic gradient and proximal-gradient methods under sub-Weibull noise, showing linear convergence up to a noise-dependent error, with bounds applicable to static and dynamic costs.

Contribution

It introduces high-probability convergence bounds for online gradient methods with sub-Weibull noise under the PL condition, extending existing results to more general noise models.

Findings

Linear convergence up to noise-dependent error

High-probability bounds derived for sub-Weibull noise

New regret bounds for static costs with sub-Weibull errors

Abstract

This paper focuses on the online gradient and proximal-gradient methods with stochastic gradient errors. In particular, we examine the performance of the online gradient descent method when the cost satisfies the Polyak-\L ojasiewicz (PL) inequality. We provide bounds in expectation and in high probability (that hold iteration-wise), with the latter derived by leveraging a sub-Weibull model for the errors affecting the gradient. The convergence results show that the instantaneous regret converges linearly up to an error that depends on the variability of the problem and the statistics of the sub-Weibull gradient error. Similar convergence results are then provided for the online proximal-gradient method, under the assumption that the composite cost satisfies the proximal-PL condition. In the case of static costs, we provide new bounds for the regret incurred by these methods when the gradient errors are modeled as sub-Weibull random variables. Illustrative simulations are provided to corroborate the technical findings.