AI Paper Atlas

Research map / Learning theory and algorithms / Optimal transport

Geometry-Misalignment in Distributional Learning

Tao Wang, Xiaoting Zhong

ICML 2026 · Learning theory and algorithms › Optimal transport · approach: Neural optimal transport

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 geometry-aware preconditioned updates instead of Euclidean SGDSinkhorn distribution matching · DGP-1 high misalignmentiterations to converge: 1600 -> 130Improved
add misalignment gating (GCO) to always-on geometry-aware updates (ablation)geometry-aware optimization · DGP-1 high misalignmentwall-clock time: 3.30 -> 1.70 (iterations 130 -> 145)Improved
add gating to GCO (versus no-gate variant) (ablation)GCO · DGP-1 low misalignmentwall-clock time: 2.60 -> 1.50Improved
add gating to GCO (ablation)GCO · DGP-1 high misalignmentiterations / objective gap: iterations 135 -> 145; gap 7.1e-5 -> 7.5e-5Hurt
use Shampoo or K-FAC instead of SGDEuclidean first-order baseline · DGP-1 high misalignmentiterations and overhead: iterations 1600 -> 720 (Shampoo), 690 (K-FAC); overhead 1.27 and 1.43 vs 1.00Mixed
use Adam or AdaGrad instead of SGDEuclidean first-order baseline · DGP-1 high misalignmentiterations: 1600 -> 1450 (Adam), 1700 (AdaGrad)No clear effect
increase geometry-misalignment from low to high (ablation)SGD · DGP-1 Linear Anisotropyiterations to converge: 180 -> 1600Hurt
increase misalignment from low to high (ablation)GCO · DGP-1 Linear Anisotropyfraction of geometry-aware updates (%): 6 -> 34Improved
use GCO instead of domain adaptation baselines (CORAL, DANN, CDAN, MMD)distribution matching · DGP-1 high misalignmentobjective gap: 1.1 × 10−3 (CDAN) -> 7.5 × 10−5Improved

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