Research map / Learning theory and algorithms / Optimal transport
Fast Approximation of the Sliced-Wasserstein Distance Using Concentration of Random Projections
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 |
|---|---|---|---|
| regularize the discriminator (Reg-SWG vs SWG) (ablation) | SWG · MNIST | FID: 22.41 -> 15.53 | Helped |
| regularize the discriminator (Reg-SWG vs SWG) (ablation) | SWG · CelebA | FID: 31.04 -> 24.14 | Helped |
| replace Monte Carlo SW with approximate SW (Reg-det-SWG vs Reg-SWG) (ablation) | Reg-SWG · MNIST | FID: 15.53 -> 15.72 | No clear effect |
| replace Monte Carlo SW with approximate SW (Reg-det-SWG vs Reg-SWG) (ablation) | Reg-SWG · CelebA | FID: 24.14 -> 23.65 | No clear effect |
| replace Monte Carlo SW with approximate SW on GPU (ablation) | Reg-SWG · MNIST | T_SW (s/epoch, GPU): 1.1 -> 0.07 | Helped |
| replace Monte Carlo SW with approximate SW on GPU (ablation) | Reg-SWG · CelebA | T_SW (s/epoch, GPU): 10.0 -> 1.3 | Helped |
| replace Monte Carlo SW with approximate SW on CPU (ablation) | Reg-SWG · CelebA | T_SW (s/epoch, CPU): 2.7 $\times 10^3$ -> 2.6 | Helped |
| replace Monte Carlo SW with approximate SW on CPU (ablation) | Reg-SWG · MNIST | T_SW (s/epoch, CPU): 1.1 $\times 10^2$ -> 0.2 | Helped |
| regularize the discriminator (ablation) | SWG · MNIST | T_tot (s/epoch, GPU): 4.5 -> 6.5 | Hurt |
| regularize the discriminator (ablation) | SWG · CelebA | T_tot (s/epoch, GPU): 3.9 $\times 10^2$ -> 4.4 $\times 10^2$ | Hurt |
| approximate SW instead of Monte Carlo L=5000 | Monte Carlo SW · Gamma synthetic d=10^3 | computation time: 8s -> 0.02s | Helped |
| decrease number of projections L in Monte Carlo (ablation) | Monte Carlo SW · Gamma synthetic | approximation error | Hurt |
Closest papers by meaning
- Sliced Wasserstein Estimation with Control Variates (ICLR 2024 · similarity 0.91)
- Statistical and Computational Guarantees of Kernel Max-Sliced Wasserstein Distances (ICML 2025 · similarity 0.89)
- Sliced Wasserstein with Random-Path Projecting Directions (ICML 2024 · similarity 0.89)
- Tree-Sliced Wasserstein Distance with Nonlinear Projection (ICML 2025 · similarity 0.89)
- Sliced-Wasserstein Estimation with Spherical Harmonics as Control Variates (ICML 2024 · similarity 0.89)
- Towards Understanding Gradient Dynamics of the Sliced-Wasserstein Distance via Critical Point Analysis (ICML 2025 · similarity 0.88)
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.