AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Variational Inference for Uncertain Optimal Transport via Sinkhorn Parametrization

Ananyapam De, Linus Bleistein, Anton Frederik Thielmann, Benjamin Säfken

ICML 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Annealing and divergence-based variational objectives

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 SPVI instead of HMC samplingBayesOT posterior inference · Gaussian point clouds, n=256runtime (s): 21,533 -> 2,144Improved
use SPVI instead of HMC samplingBayesOT posterior inference · Gaussian point clouds, n=128runtime (s): 1,217.1 -> 580Improved
use SPVI instead of HMC samplingBayesOT posterior inference · Gaussian point clouds, n=64runtime (s): 329.0 -> 157Improved
use SPVI instead of HMC samplingBayesOT posterior inference · Gaussian point clouds, n=32runtime (s): 61.6 -> 59No clear effect
use SPVI instead of HMC samplingBayesOT posterior inference · Gaussian point clouds, n=512runtime (s)Improved
use SPVI instead of MAP on mean costSinkhorn MAP · Gaussian point clouds, n=512runtime (s): 0.161 -> 7,933Hurt
use SPVI instead of source-only BaseBase classifier · Office-31 (mean of 6 tasks)accuracy (%): 87.18 -> 90.77Improved
use SPVI instead of JDOTJDOT · Office-31 (mean of 6 tasks)accuracy (%): 90.84 -> 90.77No clear effect
use SPVI instead of OT-MMOT-MM · Office-31 (mean of 6 tasks)accuracy (%): 90.48 -> 90.77No clear effect
use SPVI instead of SurKSurK · Office-31 (mean of 6 tasks)accuracy (%): 86.27 -> 90.77Improved
use SPVI instead of JDOTJDOT · Office-31 amazon->dslraccuracy (%): 87.48 -> 86.81No clear effect

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