AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

"On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond"

Yuzhe Yang, Hao Wang, Dina Katabi

ECCV 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Optimization-based domain generalization

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
add BoDA loss (decoupled classifier)ERM · MDLT average over five datasetsaccuracy: 77.6 -> 80.2Improved
add decoupled classifier learning stage with class-balanced sampling (ablation)BoDA · MDLT average over five datasetsaccuracy: 78.8 -> 80.2Improved
add momentum to BoDA (ablation)BoDA_r,c · MDLT average over five datasetsaccuracy: 80.2 -> 80.5Improved
use balanced distance (BoDA instead of vanilla DA) (ablation)DA loss · MDLT datasets averageaccuracy: 2.3% on averageImproved
vary distance calibration coefficient lambda (ablation)BoDA · MDLT datasets averageaccuracy: 1.9% to 2.9% over ERMNo clear effect
apply BoDA on severe-imbalance datasetERM · TerraInc-MLTaverage accuracy: 75.3 -> 83.0Improved
apply BoDA on balanced-ish datasetERM · PACS-MLTaverage accuracy: 97.1 -> 97.2No clear effect
apply BoDA on few-shot pairsERM · PACS-MLT few-shot pairsaccuracy: 98.0 -> 96.0Hurt
apply BoDA to zero-shot pairsERM · VLCS-MLT zero-shot pairsaccuracy: 32.9 -> 43.3Improved
add BoDA to domain generalizationERM · DG average over five datasetsaccuracy: 63.3 -> 65.5Improved
combine BoDA with current DG SOTACurrent SOTA · DG average over five datasetsaccuracy: 64.5 -> 66.2Improved
use BoDA over TerraInc DGERM · TerraInc (DG)accuracy: 46.1 -> 50.2Improved

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