On the Nash equilibrium of moment-matching GANs for stationary Gaussian processes - Algorithmes Parallèles et Optimisation Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

On the Nash equilibrium of moment-matching GANs for stationary Gaussian processes

Résumé

Generative Adversarial Networks (GANs) learn an implicit generative model from data samples through a two-player game. In this paper, we study the existence of Nash equilibrium of the game which is consistent as the number of data samples grows to infinity. In a realizable setting where the goal is to estimate the ground-truth generator of a stationary Gaussian process, we show that the existence of consistent Nash equilibrium depends crucially on the choice of the discriminator family. The discriminator defined from second-order statistical moments can result in non-existence of Nash equilibrium, existence of consistent non-Nash equilibrium, or existence and uniqueness of consistent Nash equilibrium, depending on whether symmetry properties of the generator family are respected. We further study empirically the local stability and global convergence of gradient descent-ascent methods towards consistent equilibrium.

Dates et versions

hal-03736803 , version 1 (22-07-2022)

Licence

Paternité

Identifiants

Citer

Sixin Zhang. On the Nash equilibrium of moment-matching GANs for stationary Gaussian processes. Machine Learning Research 2022, Aug 2022, Beijing, China. pp.1-26, ⟨10.48550/arXiv.2203.07136⟩. ⟨hal-03736803⟩
49 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More