Finite Difference Gradient Approximation: To Randomize or Not?

DOI10.1287/ijoc.2022.1218OpenAlexW4289767001WikidataQ114058177 ScholiaQ114058177MaRDI QIDQ5057983

Katya Scheinberg

Publication date: 1 December 2022

Published in: INFORMS Journal on Computing (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1287/ijoc.2022.1218

zbMATH Keywords

finite difference approximationgradient descentrandomized

Mathematics Subject Classification ID

Mathematical programming (90Cxx)

Related Items

Gradient-free federated learning methods with \(l_1\) and \(l_2\)-randomization for non-smooth convex stochastic optimization problems

Uses Software

• Spearmint

Cites Work

• Unnamed Item
• Spearmint
• A theoretical and empirical comparison of gradient approximations in derivative-free optimization
• Random gradient-free minimization of convex functions
• Adaptive stochastic approximation by the simultaneous perturbation method
• Optimal Rates for Zero-Order Convex Optimization: The Power of Two Function Evaluations
• Estimating Derivatives of Noisy Simulations
• Multivariate stochastic approximation using a simultaneous perturbation gradient approximation
• A Derivative-Free Trust-Region Algorithm for the Optimization of Functions Smoothed via Gaussian Convolution Using Adaptive Multiple Importance Sampling
• Derivative-free optimization methods
• An Optimal Algorithm for Bandit and Zero-Order Convex Optimization with Two-Point Feedback
• Stochastic First- and Zeroth-Order Methods for Nonconvex Stochastic Programming