Adaptive and Optimal Second-order Optimistic Methods for Minimax Optimization

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Jiang, Ruichen, Kavis, Ali, Jin, Qiujiang, Sanghavi, Sujay, Mokhtari, Aryan
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866929587464699904
author Jiang, Ruichen
Kavis, Ali
Jin, Qiujiang
Sanghavi, Sujay
Mokhtari, Aryan
author_facet Jiang, Ruichen
Kavis, Ali
Jin, Qiujiang
Sanghavi, Sujay
Mokhtari, Aryan
contents We propose adaptive, line search-free second-order methods with optimal rate of convergence for solving convex-concave min-max problems. By means of an adaptive step size, our algorithms feature a simple update rule that requires solving only one linear system per iteration, eliminating the need for line search or backtracking mechanisms. Specifically, we base our algorithms on the optimistic method and appropriately combine it with second-order information. Moreover, distinct from common adaptive schemes, we define the step size recursively as a function of the gradient norm and the prediction error in the optimistic update. We first analyze a variant where the step size requires knowledge of the Lipschitz constant of the Hessian. Under the additional assumption of Lipschitz continuous gradients, we further design a parameter-free version by tracking the Hessian Lipschitz constant locally and ensuring the iterates remain bounded. We also evaluate the practical performance of our algorithm by comparing it to existing second-order algorithms for minimax optimization.
format Preprint
id arxiv_https___arxiv_org_abs_2406_02016
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Adaptive and Optimal Second-order Optimistic Methods for Minimax Optimization
Jiang, Ruichen
Kavis, Ali
Jin, Qiujiang
Sanghavi, Sujay
Mokhtari, Aryan
Optimization and Control
Machine Learning
We propose adaptive, line search-free second-order methods with optimal rate of convergence for solving convex-concave min-max problems. By means of an adaptive step size, our algorithms feature a simple update rule that requires solving only one linear system per iteration, eliminating the need for line search or backtracking mechanisms. Specifically, we base our algorithms on the optimistic method and appropriately combine it with second-order information. Moreover, distinct from common adaptive schemes, we define the step size recursively as a function of the gradient norm and the prediction error in the optimistic update. We first analyze a variant where the step size requires knowledge of the Lipschitz constant of the Hessian. Under the additional assumption of Lipschitz continuous gradients, we further design a parameter-free version by tracking the Hessian Lipschitz constant locally and ensuring the iterates remain bounded. We also evaluate the practical performance of our algorithm by comparing it to existing second-order algorithms for minimax optimization.
title Adaptive and Optimal Second-order Optimistic Methods for Minimax Optimization
topic Optimization and Control
Machine Learning
url https://arxiv.org/abs/2406.02016