Research map / Image generation / GANs
Some Theoretical Insights into Wasserstein GANs
Gérard Biau, Maxime Sangnier, Ugo Tanielian
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase discriminator depth q (ablation) | WGAN with GroupSort discriminator · synthetic Gaussian mixtures (asymptotic) | Wasserstein distance and recall | Improved |
| increase generator depth p (ablation) | WGAN with GroupSort discriminator · synthetic Gaussian mixtures (asymptotic) | Wasserstein distance and recall | Hurt |
| increase discriminator depth q (ablation) | WGAN · synthetic Gaussian mixtures (finite sample, n=5000) | maximal Wasserstein distance | Improved |
| increase discriminator depth q (ablation) | WGAN · MNIST and Fashion-MNIST | embedded Wasserstein distance | Improved |
| increase discriminator depth q from 2 to 5 (ablation) | discriminator class vs Lip1 distance · Gaussian mixtures K=1,4,9,25 | width (b-a) of monotone equivalence fit | Improved |
| increase number of mixture components K (ablation) | discriminator class vs Lip1 distance · fixed depth q | width (b-a) of monotone equivalence fit | Hurt |
Closest papers by meaning
- Wasserstein Generative Adversarial Networks (ICML 2017 · similarity 0.94)
- Improving the Improved Training of Wasserstein GANs: A Consistency Term and Its Dual Effect (ICLR 2018 · similarity 0.94)
- A Two-Step Computation of the Exact GAN Wasserstein Distance (ICML 2018 · similarity 0.93)
- Bridging the Gap Between f-GANs and Wasserstein GANs (ICML 2020 · similarity 0.93)
- Wasserstein Divergence for GANs (ECCV 2018 · similarity 0.93)
- Improved Training of Wasserstein GANs (NeurIPS 2017 · similarity 0.92)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
