Research map / Learning theory and algorithms / Optimal transport
Tree-sliced Sobolev IPM
Viet-Hoang Tran, Thanh Tran, Thanh Chu, Duy-Tung Pham, Trung-Khang Tran, Tam Le, Tan Minh Nguyen
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 |
|---|---|---|---|
| add TS-Sobolev (p=1.2) instead of Db-TSW in gradient flow | TSW · 8 Gaussians | Wasserstein distance: 2.50e-6 -> 8.88e-7 | Improved |
| add TS-Sobolev (p=1.2) instead of Db-TSW | TSW · Gaussian 30d | Wasserstein distance: 1.78 -> 1.40 | Improved |
| raise order from p=1.2 to p=2 (ablation) | TS-Sobolev · Gaussian 30d gradient flow | Wasserstein distance: 1.40 -> 3.68 | Hurt |
| raise order from p=1.2 to p=2 (ablation) | TS-Sobolev · 8 Gaussians gradient flow | Wasserstein distance: 8.88e-7 -> 3.50e-6 | Hurt |
| replace Db-TSW-DD⊥ discriminator loss with TS-Sobolev2-DD | DDGAN · CIFAR-10 | FID: 2.53 -> 2.277 | Improved |
| raise order from p=1.5 to p=2 (ablation) | TS-Sobolev-DD · CIFAR-10 | FID: 2.302 -> 2.277 | Improved |
| replace Gaussian-kernel uniformity with STS-Sobolev2 instead of STSW | SSL · CIFAR-10 projected features | accuracy: 76.78 -> 77.65 | Improved |
| raise order from p=1.5 to p=2 (ablation) | STS-Sobolev · CIFAR-10 SSL encoded features | accuracy: 79.88 -> 80.6 | Improved |
| use STS-Sobolev2 instead of STSW | spherical gradient flow · 12 vMFs epoch 250 | log Wasserstein distance: -3.549 -> -3.616 | Improved |
| use TS-Sobolev2 instead of TSW-SL | topic model · BBC | topic coherence: 0.796 -> 0.805 | Improved |
| use STS-Sobolev2 instead of STSW | topic model · M10 | topic coherence: 0.373 -> 0.423 | Improved |
| increase number of trees L (ablation) | TS-Sobolev · Monte Carlo approximation | approximation error | Improved |
Closest papers by meaning
- Tree-Sliced Wasserstein Distance with Nonlinear Projection (ICML 2025 · similarity 0.92)
- Tree-Sliced Wasserstein Distance: A Geometric Perspective (ICML 2025 · similarity 0.91)
- Distance-Based Tree-Sliced Wasserstein Distance (ICLR 2025 · similarity 0.90)
- Spherical Tree-Sliced Wasserstein Distance (ICLR 2025 · similarity 0.90)
- Towards Understanding Gradient Dynamics of the Sliced-Wasserstein Distance via Critical Point Analysis (ICML 2025 · similarity 0.88)
- Statistical and Computational Guarantees of Kernel Max-Sliced Wasserstein Distances (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.
