Zeroth-order algorithms for stochastic distributed nonconvex optimization

Xinlei Yi

doi:10.1016/j.automatica.2022.110353

Return

Zeroth-order algorithms for stochastic distributed nonconvex optimization

2022-08-01

Xinlei Yi

Shengjun Zhang

Tao Yang *

Karl Henrik Johansson

DOI：10.1016/j.automatica.2022.110353

Original

Save

View PDF

Abstract

En 中文

In this paper, we consider a stochastic distributed nonconvex optimization problem with the cost function being distributed over n agents having access only to zeroth-order (ZO) information of the cost. This problem has various machine learning applications. As a solution, we propose two distributed ZO algorithms, in which at each iteration each agent samples the local stochastic ZO oracle at two points with a time-varying smoothing parameter. We show that the proposed algorithms achieve the linear speedup convergence rate O(root p/(nT)) for smooth cost functions under the state-dependent variance assumptions which are more general than the commonly used bounded variance and Lipschitz assumptions, and O(p/(nT)) convergence rate when the global cost function additionally satisfies the Polyak-Lojasiewicz (P-L) condition, where p and T are the dimension of the decision variable and the total number of iterations, respectively. To the best of our knowledge, this is the first linear speedup result for distributed ZO algorithms. It consequently enables systematic processing performance improvements by adding more agents. We also show that the proposed algorithms converge linearly under the relatively bounded second moment assumptions and the P-L condition. We demonstrate through numerical experiments the efficiency of our algorithms on generating adversarial examples from deep neural networks in comparison with baseline and recently proposed centralized and distributed ZO algorithms. (C) 2022 Elsevier Ltd. All rights reserved.

Keywords:

Distributed nonconvex optimization

Gradient-free

Linear speedup

Polyak-Lojasiewicz condition

Stochastic optimization

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.