Research map / Learning theory and algorithms / Optimal transport
A Riemannian Exponential Augmented Lagrangian Method for Computing the Projection Robust Wasserstein Distance
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 |
|---|---|---|---|
| replace RBCD with iRBBS-0.1 (inexact Riemannian BB + Sinkhorn) | subproblem (11) solver · Shakespeare operas | running time (s): 6.89 -> 0.66 | Helped |
| replace RABCD with iRBBS-0.1 | subproblem (11) solver · Shakespeare operas | running time (s): 4.62 -> 0.66 | Helped |
| use adaptive BB stepsize instead of hand-tuned stepsize | iRBBS vs R(A)BCD · Shakespeare operas | need for stepsize tuning | Helped |
| iRBBS-0.1 vs RBCD | subproblem (11) · Shakespeare operas | nGrad: 1000 -> 63/1460 | Helped |
| set theta=0.1 instead of theta=inf (one Sinkhorn step) (ablation) | iRBBS · Shakespeare operas | running time (s): 1.86 -> 0.66 | Helped |
| set theta=0.1 instead of theta=0 (near-exact gradient) (ablation) | iRBBS · Shakespeare operas | running time (s): 0.90 -> 0.66 | Helped |
| theta=0.1 instead of theta=inf (ablation) | iRBBS · Shakespeare operas | solution quality P2k: 0.075387 -> 0.075757 | Helped |
| theta=0 (more accurate gradient) vs theta=0.1 (ablation) | iRBBS · Shakespeare operas | nSk: 65/3200 vs 63/1460 | Hurt |
| update multiplier matrix (gamma_W=0.9) with larger eta_min 0.007 (ablation) | REALM · Shakespeare operas | running time (s): 78.0 -> 15.0 | Helped |
| update multiplier matrix (gamma_W=0.9) with larger eta_min 0.007 (ablation) | REALM · Shakespeare operas | solution quality P2k: 0.11338 -> 0.11461 | Helped |
| update multiplier matrix (gamma_W=0.9) (ablation) | REALM · Shakespeare operas | nGrad: 863 -> 787 | Helped |
| update multiplier matrix (ablation) | REALM · Shakespeare MV/O pair | solution quality P2k: 0.04181 -> 0.03661 | Hurt |
Closest papers by meaning
- Acceleration via silver step-size on Riemannian manifolds with applications to Wasserstein space (NeurIPS 2025 · similarity 0.85)
- Riemannian Accelerated Zeroth-order Algorithm: Improved Robustness and Lower Query Complexity (ICML 2024 · similarity 0.85)
- Tree-Sliced Wasserstein Distance with Nonlinear Projection (ICML 2025 · similarity 0.85)
- Continuous-time Riemannian SGD and SVRG Flows on Wasserstein Probabilistic Space (NeurIPS 2025 · similarity 0.85)
- Efficient Optimization with Orthogonality Constraint: a Randomized Riemannian Submanifold Method (ICML 2025 · similarity 0.85)
- Implicit Riemannian Optimism with Applications to Min-Max Problems (ICML 2025 · similarity 0.84)
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.