Optimal 1-Wasserstein distance for WGANs - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Bernoulli Année : 2024

Optimal 1-Wasserstein distance for WGANs

Arthur Stéphanovitch
  • Fonction : Auteur
  • PersonId : 1288905
Ugo Tanielian
  • Fonction : Auteur
  • PersonId : 1288906
Benoît Cadre
Gérard Biau

Résumé

The mathematical forces at work behind Generative Adversarial Networks raise challenging theoretical issues. Motivated by the important question of characterizing the geometrical properties of the generated distributions, we provide a thorough analysis of Wasserstein GANs (WGANs) in both the finite sample and asymptotic regimes. We study the specific case where the latent space is univariate and derive results valid regardless of the dimension of the output space. We show in particular that for a fixed sample size, the optimal WGANs are closely linked with connected paths minimizing the sum of the squared Euclidean distances between the sample points. We also highlight the fact that WGANs are able to approach (for the 1-Wasserstein distance) the target distribution as the sample size tends to infinity, at a given convergence rate and provided the family of generative Lipschitz functions grows appropriately. We derive in passing new results on optimal transport theory in the semi-discrete setting.
Fichier principal
Vignette du fichier
Optimal_1_Wasserstein_distance_for_WGANs.pdf (2.06 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04223292 , version 1 (29-09-2023)

Identifiants

  • HAL Id : hal-04223292 , version 1

Citer

Arthur Stéphanovitch, Ugo Tanielian, Benoît Cadre, Nicolas Klutchnikoff, Gérard Biau. Optimal 1-Wasserstein distance for WGANs. Bernoulli, In press. ⟨hal-04223292⟩
21 Consultations
4 Téléchargements

Partager

Gmail Facebook X LinkedIn More