Research map / Learning theory and algorithms / Optimal transport
A Two-Step Computation of the Exact GAN Wasserstein Distance
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 weight clipping/penalty critic with two-step LP + DNN regression (WGAN-TS) | WGAN critic · 8 Gaussian toy | accuracy of approximated Wasserstein distance | Helped |
| Replace WGAN-GP critic penalty with two-step LP + DNN regression | WGAN-GP critic · CIFAR-10 | Inception score: +0.14 | Helped |
| Replace WGAN-GP critic penalty with two-step LP + DNN regression | WGAN-GP critic · MNIST | Inception score: +0.01 | No clear effect |
| Replace WGAN-GP critic with two-step method | critic time per generator iteration · CIFAR-10 | critic time (s): 0.881 to 0.278 | Helped |
| Use two-step method instead of weight clipping | WGAN · MNIST | Inception score: +0.71 | Helped |
| Use two-step method instead of weight clipping | WGAN · CIFAR-10 | Inception score: +0.36 | Helped |
| Weight scaling of critic after each update to satisfy constraints | WGAN-TS · LSUN | fraction of beta equal to 1 | No clear effect |
Closest papers by meaning
- Banach Wasserstein GAN (NeurIPS 2018 · similarity 0.94)
- Towards Generalized Implementation of Wasserstein Distance in GANs (AAAI 2021 · similarity 0.93)
- Wasserstein Generative Adversarial Networks (ICML 2017 · similarity 0.93)
- Generative Modeling Using the Sliced Wasserstein Distance (CVPR 2018 · similarity 0.93)
- Improving the Improved Training of Wasserstein GANs: A Consistency Term and Its Dual Effect (ICLR 2018 · similarity 0.92)
- Do WGANs succeed because they minimize the Wasserstein Distance? Lessons from Discrete Generators (ICLR 2025 · 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.