AI Paper Atlas

Research map / Learning theory and algorithms / Optimal transport

Online Sinkhorn: Optimal Transport distances from sample streams

Arthur Mensch, Gabriel Peyré

NeurIPS 2020 · Learning theory and algorithms › Optimal transport · approach: Transport map estimation · 3 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use Online Sinkhorn instead of batch SinkhornSinkhorn potential estimation · Gaussian mixtures, 1D and 2Dpotential estimation errorImproved
Use Online Sinkhorn instead of batch SinkhornSinkhorn potential estimation · Gaussian mixtures, 10Dpotential estimation errorMixed
Use fully-corrective Online Sinkhorn (ablation)Online Sinkhorn · Gaussian mixtures, 10Dpotential estimation errorImproved
Use Online Sinkhorn instead of SinkhornSinkhorn potential estimation · Gaussian mixtures, 1D and 2D, N=10000potential estimation errorNo clear effect
Use randomized Sinkhorn instead of batch Sinkhorn (ablation)Sinkhorn potential estimation · Gaussian mixturespotential estimation errorNo clear effect
Represent potentials with Online Sinkhorn's kernel expansion instead of an RKHS expansion trained by SGDPotential estimation · Gaussian mixtures, 1D and 2Dpotential estimation error in low-mass regionsImproved
Use Online Sinkhorn instead of SGD with an RKHS parametrizationPotential estimation · Gaussian mixturesstabilityImproved
Warm up Sinkhorn with Online Sinkhorn before standard Sinkhorn updatesSinkhorn · Discrete GMMs, 3D dragon, and spheredistance to optimumImproved
Warm up Sinkhorn with Online Sinkhorn before standard Sinkhorn updatesSinkhorn · Discrete GMMs, 3D dragon, and sphereconvergence speedImproved
Reduce ε while using Online Sinkhorn warmup (ablation)OS-warmed-up Sinkhorn · Discrete OT datasetsspeed gainImproved
Sample without replacement during warmup (ablation)OS-warmed-up Sinkhorn · Discrete OT datasetsconvergence speedImproved
Use fixed batch sizes instead of increasing batch sizes (ablation)Online Sinkhorn · Continuous OTpotential convergenceHurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy