ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2405.17761
32
1

Double Variance Reduction: A Smoothing Trick for Composite Optimization Problems without First-Order Gradient

28 May 2024
Hao Di
Haishan Ye
Yueling Zhang
Xiangyu Chang
Guang Dai
Ivor W. Tsang
ArXivPDFHTML
Abstract

Variance reduction techniques are designed to decrease the sampling variance, thereby accelerating convergence rates of first-order (FO) and zeroth-order (ZO) optimization methods. However, in composite optimization problems, ZO methods encounter an additional variance called the coordinate-wise variance, which stems from the random gradient estimation. To reduce this variance, prior works require estimating all partial derivatives, essentially approximating FO information. This approach demands O(d) function evaluations (d is the dimension size), which incurs substantial computational costs and is prohibitive in high-dimensional scenarios. This paper proposes the Zeroth-order Proximal Double Variance Reduction (ZPDVR) method, which utilizes the averaging trick to reduce both sampling and coordinate-wise variances. Compared to prior methods, ZPDVR relies solely on random gradient estimates, calls the stochastic zeroth-order oracle (SZO) in expectation O(1)\mathcal{O}(1)O(1) times per iteration, and achieves the optimal O(d(n+κ)log⁡(1ϵ))\mathcal{O}(d(n + \kappa)\log (\frac{1}{\epsilon}))O(d(n+κ)log(ϵ1​)) SZO query complexity in the strongly convex and smooth setting, where κ\kappaκ represents the condition number and ϵ\epsilonϵ is the desired accuracy. Empirical results validate ZPDVR's linear convergence and demonstrate its superior performance over other related methods.

View on arXiv
Comments on this paper