Research map / Image generation / GANs
Improved Training of Wasserstein GANs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace critic weight clipping with a gradient penalty | WGAN · CIFAR-10 | Inception score | Helped |
| Use WGAN-GP instead of the standard GAN objective | Random GAN architectures · ImageNet 32×32, minimum score 5.0 | architectures trained exclusively: Only GAN: 0; Only WGAN-GP: 147 | Helped |
| Use WGAN-GP instead of the standard GAN objective | Random GAN architectures · ImageNet 32×32, minimum score 7.0 | architectures trained exclusively: Only GAN: 1; Only WGAN-GP: 104 | Helped |
| Use WGAN-GP ResNet instead of SGAN | Conditional CIFAR-10 generation · CIFAR-10 supervised | Inception score: 8.59 ± .12 -> 8.42 ± .10 | Hurt |
Closest papers by meaning
- Improving the Improved Training of Wasserstein GANs: A Consistency Term and Its Dual Effect (ICLR 2018 · similarity 0.93)
- Wasserstein Generative Adversarial Networks (ICML 2017 · similarity 0.93)
- On the regularization of Wasserstein GANs (ICLR 2018 · similarity 0.93)
- A Two-Step Computation of the Exact GAN Wasserstein Distance (ICML 2018 · similarity 0.91)
- Generative Modeling Using the Sliced Wasserstein Distance (CVPR 2018 · similarity 0.91)
- Lipschitz Generative Adversarial Nets (ICML 2019 · similarity 0.91)
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.