Minimax Efficient Finite-Difference Stochastic Gradient Estimators Using Black-Box Function Evaluations

DOI10.1016/J.ORL.2020.10.013MaRDI QIDQ6344711arXivFDO

Authors Henry Lam, Haidong Li, Xuhui Zhang

Publication date 8 July 2020

Published in arXiv

Full work available at URL https://doi.org/10.1016/J.ORL.2020.10.013

Mathematics Subject Classification ID

Minimax procedures in statistical decision theory (62C20) Nonparametric estimation (62G05) Stochastic programming (90C15) Minimax problems in mathematical programming (90C47) Derivative-free methods and methods using generalized derivatives (90C56)

arXiv classification

math.STstat.MEstat.TH

Abstract: Standard approaches to stochastic gradient estimation, with only noisy black-box function evaluations, use the finite-difference method or its variants. While natural, it is open to our knowledge whether their statistical accuracy is the best possible. This paper argues so by showing that central finite-difference is a nearly minimax optimal zeroth-order gradient estimator for a suitable class of objective functions and mean squared risk, among both the class of linear estimators and the much larger class of all (nonlinear) estimators.

This page was built for publication: Minimax Efficient Finite-Difference Stochastic Gradient Estimators Using Black-Box Function Evaluations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6344711)