Escaping limit cycles: Global convergence for constrained nonconvex-nonconcave minimax problems - Département Image, Données, Signal Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Escaping limit cycles: Global convergence for constrained nonconvex-nonconcave minimax problems

Résumé

This paper introduces a new extragradient-type algorithm for a class of nonconvex-nonconcave minimax problems. It is well-known that finding a local solution for general minimax problems is computationally intractable. This observation has recently motivated the study of structures sufficient for convergence of first order methods in the more general setting of variational inequalities when the so-called weak Minty variational inequality (MVI) holds. This problem class captures non-trivial structures as we demonstrate with examples, for which a large family of existing algorithms provably converge to limit cycles. Our results require a less restrictive parameter range in the weak MVI compared to what is previously known, thus extending the applicability of our scheme. The proposed algorithm is applicable to constrained and regularized problems, and involves an adaptive stepsize allowing for potentially larger stepsizes. Our scheme also converges globally even in settings where the underlying operator exhibits limit cycles. Moreover, a variant with stochastic oracles is proposed-making it directly relevant for training of generative adversarial networks. For the stochastic algorithm only one of the stepsizes is required to be diminishing while the other may remain constant, making it interesting even in the monotone setting.
Fichier principal
Vignette du fichier
escaping_limit_cycles_global_c.pdf (2.43 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03602455 , version 1 (09-03-2022)
hal-03602455 , version 2 (14-06-2022)

Identifiants

  • HAL Id : hal-03602455 , version 1

Citer

Thomas Pethick, Panagiotis Patrinos, Olivier Fercoq, Volkan Cevherå, Puya Latafat. Escaping limit cycles: Global convergence for constrained nonconvex-nonconcave minimax problems. International Conference on Learning Representations, 2022, online, France. ⟨hal-03602455v1⟩
258 Consultations
165 Téléchargements

Partager

Gmail Facebook X LinkedIn More