Research map / Learning theory and algorithms / Optimal transport
Massively scalable Sinkhorn distances via the Nyström method
Jason Altschuler, Francis Bach, Alessandro Rudi, Jonathan Niles-Weed
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 Nystrom low-rank kernel approximation with Sinkhorn scaling instead of GPU Dual-Sink with annealing | Nys-Sink · Armadillo vs dragon (n≈3e5) | time (s): 35.4 -> 0.4 | Improved |
| use Nystrom low-rank kernel approximation with Sinkhorn scaling instead of multiscale annealing Dual-Sink | Nys-Sink · Armadillo vs xyz-dragon (n≈3.8e6) | time (s): 103.6 -> 6.3 | Improved |
| use Nys-Sink instead of standard Sinkhorn on random point clouds n≈20000 | Nys-Sink · Random point clouds | time to reach accuracy | Improved |
| increase ambient dimension from 5 to 100 with fixed rank r=300 (ablation) | Nys-Sink · 1-D curves embedded in R^d | approximation error: error smaller than 10^-4 | No clear effect |
Closest papers by meaning
- Accelerating Sinkhorn algorithm with sparse Newton iterations (ICLR 2024 · similarity 0.90)
- Faster Wasserstein Distance Estimation with the Sinkhorn Divergence (NeurIPS 2020 · similarity 0.90)
- Differential Properties of Sinkhorn Approximation for Learning with Wasserstein Distance (NeurIPS 2018 · similarity 0.89)
- Debiased Sinkhorn barycenters (ICML 2020 · similarity 0.89)
- Wasserstein Wormhole: Scalable Optimal Transport Distance with Transformer (ICML 2024 · similarity 0.85)
- Scalable Approximation Algorithms for $p$-Wasserstein Distance and Its Variants (ICML 2025 · similarity 0.85)
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.
