Research map / Learning theory and algorithms / Optimal transport
Sliced Optimal Partial Transport
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 |
|---|---|---|---|
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of ICP-D | Point cloud registration · Noisy point clouds, 10k,5% | Frobenius error: ICP-D: 1.10(1.59); Ours: 0.01(1e-3) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of ICP-D | Point cloud registration · Noisy point clouds, 10k,7% | Frobenius error: ICP-D: 3.60(0.11); Ours: 0.02(2e-3) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of ICP-D | Point cloud registration · Noisy point clouds, 9k,5% | Frobenius error: ICP-D: 1.65 (1.13); Ours: 0.20(0.09) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of ICP-D | Point cloud registration · Noisy point clouds, 9k,7% | Frobenius error: ICP-D: 2.04 (2.28); Ours: 0.33 (0.03) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of SPOT | Point cloud registration · Noisy point clouds, 10k,5% | Frobenius error: SPOT: 1.27(0.01); Ours: 0.01(1e-3) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of SPOT | Point cloud registration · Noisy point clouds, 10k,7% | Frobenius error: SPOT: 1.40(0.15); Ours: 0.02(2e-3) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of SPOT | Point cloud registration · Noisy point clouds, 9k,5% | Frobenius error: SPOT: 1.42 (0.13); Ours: 0.20(0.09) | Helped |
| Use SOPT-based correspondence estimation with prior knowledge of the clean-point count instead of SPOT | Point cloud registration · Noisy point clouds, 9k,7% | Frobenius error: SPOT: 1.53 (1.50); Ours: 0.33 (0.03) | Helped |
| Use a small transport penalty, lambda < 2.0, instead of lambda = 10.0 (ablation) | SOPT color adaptation · Color adaptation | Source-image preservation | Helped |
| Use a small transport penalty, lambda < 2.0, instead of lambda = 10.0 (ablation) | SOPT color adaptation · Color adaptation | Amount of source pixels transported | Hurt |
| Use SOPT-based point cloud registration instead of ICP | Point cloud registration · Bunny, Dragon, Mumble, and Castle with 5% or 7% noise | Total convergence time: ICP: 100-260 seconds; our method: 500-700 seconds | Mixed |
| Use SOPT-based color adaptation instead of OT-based adaptation | Color adaptation · Image color transfer | Runtime: OT-based: 40-50 seconds; our method: 60-80 seconds | Hurt |
Closest papers by meaning
- Expected Sliced Transport Plans (ICLR 2025 · similarity 0.90)
- Efficient Transferable Optimal Transport via Min-Sliced Transport Plans (NeurIPS 2026 · similarity 0.90)
- Linear Spherical Sliced Optimal Transport: A Fast Metric for Comparing Spherical Data (ICLR 2025 · similarity 0.89)
- Differentiable Generalized Sliced Wasserstein Plans (NeurIPS 2025 · similarity 0.88)
- On Partial Optimal Transport: Revising the Infeasibility of Sinkhorn and Efficient Gradient Methods (AAAI 2024 · similarity 0.88)
- Amortized Optimal Transport from Sliced Potentials (NeurIPS 2026 · similarity 0.86)
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.