Simultaneous perturbation stochastic approximation: towards one-measurement per iteration

DOI10.1007/S11075-023-01528-7zbMATH Open1530.65065arXiv2203.03075MaRDI QIDQ6076930FDOQ6076930

Authors: Shiru Li, Yong Xia, Zi Xu 

Publication date: 17 October 2023

Published in: Numerical Algorithms (Search for Journal in Brave)

Abstract: When measuring the value of a function to be minimized is not only expensive but also with noise, the popular simultaneous perturbation stochastic approximation (SPSA) algorithm requires only two function values in each iteration. In this paper, we propose a method requiring only one function measurement value per iteration in the average sense. We prove the strong convergence and asymptotic normality of the new algorithm. Experimental results show the effectiveness and potential of our algorithm.

Full work available at URL: https://arxiv.org/abs/2203.03075

Recommendations

• A one-measurement form of simultaneous perturbation stochastic approximation
• Adaptive stochastic approximation by the simultaneous perturbation method
• Optimal random perturbations for stochastic approximation using a simultaneous perturbation gradient approximation
• Multivariate stochastic approximation using a simultaneous perturbation gradient approximation
• Simultaneous perturbation stochastic approximation of nonsmooth functions

zbMATH Keywords

unconstrained optimizationstochastic algorithmSPSAapproximating gradient

Mathematics Subject Classification ID

Numerical optimization and variational techniques (65K10) Stochastic approximation (62L20) Nonlinear programming (90C30) Problems with incomplete information (optimization) (49N30)

Cites Work

• Testing Unconstrained Optimization Software
• Stochastic approximation methods for constrained and unconstrained systems
• A Stochastic Approximation Method
• Title not available (Why is that?)
• On Asymptotic Normality in Stochastic Approximation
• Multivariate stochastic approximation using a simultaneous perturbation gradient approximation
• Stochastic Estimation of the Maximum of a Regression Function
• A one-measurement form of simultaneous perturbation stochastic approximation
• A stochastic approximation frame algorithm with adaptive directions
• Optimal random perturbations for stochastic approximation using a simultaneous perturbation gradient approximation
• Simultaneous perturbation stochastic approximation of nonsmooth functions
• A modified second‐order SPSA optimization algorithm for finite samples
• Statistical inferences for termination of Markov type random search algorithms

Cited In (1)

• Two-timescale simultaneous perturbation stochastic approximation using deterministic perturbation sequences

This page was built for publication: Simultaneous perturbation stochastic approximation: towards one-measurement per iteration

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6076930)