Research map / Learning theory and algorithms / Optimal transport
Hilbert Curve Projection Distance for Distribution Comparison
Tao Li, Cheng Meng, Hongteng Xu, Jun Yang Yu
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 |
|---|---|---|---|
| use Hilbert curve projection instead of sliced Wasserstein | KNN classifier · ModelNet10 (n=100) | accuracy (%): 71.5 -> 73.3 | Improved |
| use HCP instead of SW | KNN classifier · ModelNet10 (n=100) | CPU time (s): 123.9 -> 17.9 | Improved |
| use HCP instead of SW | KNN classifier · ModelNet10 (n=2000) | accuracy (%): 80.5 -> 82.3 | Improved |
| use HCP instead of Sinkhorn | KNN classifier · ModelNet10 (n=100) | accuracy (%): 77.0 -> 73.3 | Hurt |
| use HCP instead of Sinkhorn | KNN classifier · ModelNet10 (n=500) | accuracy (%): 81.5 -> 81.8 | No clear effect |
| use PRHCP instead of SW | KNN classifier · TWITTER document classification | accuracy (%): 68.6 -> 74.6 | Improved |
| use PRHCP instead of TSW | KNN classifier · TWITTER document classification | accuracy (%): 71.2 -> 74.6 | Improved |
| use HCP instead of SW for color transfer flow | Wasserstein flow · Color transfer | time (s): 496.7 -> 57.3 | Improved |
| use IPRHCP-AE instead of WAE | Autoencoder · MNIST image generation | FID: 54.61 -> 40.03 | Improved |
| use IPRHCP-AE instead of SWAE | Autoencoder · CelebA image generation | FID: 84.52 -> 56.00 | Improved |
| use PRHCP-AE instead of IPRHCP-AE | Autoencoder · CelebA image generation | FID: 56.00 -> 67.82 | Hurt |
| use HCP in Wasserstein flow instead of SW and variants | Wasserstein flow · Synthetic Circle, Swiss Roll, 25-Gaussian, Puma | W2 distance | Improved |
Closest papers by meaning
- Intrinsic Sliced Wasserstein Distances for Comparing Collections of Probability Distributions on Manifolds and Graphs (ICML 2023 · similarity 0.85)
- Statistical and Computational Guarantees of Kernel Max-Sliced Wasserstein Distances (ICML 2025 · similarity 0.85)
- Distributional Convergence of the Sliced Wasserstein Process (NeurIPS 2022 · similarity 0.85)
- A New Robust Partial p-Wasserstein-Based Metric for Comparing Distributions (ICML 2024 · similarity 0.84)
- Hierarchical Integral Probability Metrics: A distance on random probability measures with low sample complexity (ICML 2024 · similarity 0.84)
- Tree-Sliced Wasserstein Distance with Nonlinear Projection (ICML 2025 · similarity 0.83)
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.
