pith. sign in

arxiv: 2103.01678 · v4 · pith:VY3ZOIAEnew · submitted 2021-03-02 · 📊 stat.ML · cs.LG

Wasserstein GANs Work Because They Fail (to Approximate the Wasserstein Distance)

classification 📊 stat.ML cs.LG
keywords wassersteindistancegansapproximatelosstheoreticalworkanalysis
0
0 comments X
read the original abstract

Wasserstein GANs are based on the idea of minimising the Wasserstein distance between a real and a generated distribution. We provide an in-depth mathematical analysis of differences between the theoretical setup and the reality of training Wasserstein GANs. In this work, we gather both theoretical and empirical evidence that the WGAN loss is not a meaningful approximation of the Wasserstein distance. Moreover, we argue that the Wasserstein distance is not even a desirable loss function for deep generative models, and conclude that the success of Wasserstein GANs can in truth be attributed to a failure to approximate the Wasserstein distance.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Ensemble Distributionally Robust Bayesian Optimisation with Continuous Context

    cs.LG 2026-05 unverdicted novelty 6.0

    A tractable ensemble distributionally robust Bayesian optimization method achieves improved sublinear regret bounds under context uncertainty.

  2. Ensemble Distributionally Robust Bayesian Optimisation with Continuous Context

    cs.LG 2026-05 unverdicted novelty 6.0

    EDRBO uses ensemble surrogates and Wasserstein ambiguity sets to robustify BO acquisition functions against context distribution mismatch, with sublinear regret O(γ_T √T) and SOTA empirical results on continuous contexts.

  3. Learning to Emulate Chaos: Adversarial Optimal Transport Regularization

    stat.ML 2026-04 conditional novelty 6.0

    Adversarial optimal transport objectives jointly learn summary statistics and a chaotic-system emulator from a single noisy trajectory, improving long-term statistical fidelity over handcrafted-feature baselines.

  4. Learning to Emulate Chaos: Adversarial Optimal Transport Regularization

    stat.ML 2026-04 unverdicted novelty 5.0

    Adversarial optimal transport objectives train neural emulators with improved long-term statistical fidelity on chaotic systems.