AI Paper Atlas

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

Provable Domain Generalization via Invariant-Feature Subspace Recovery

Haoxiang Wang, Haozhe Si, Bo Li, Han Zhao

ICML 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Optimization-based domain generalization · 2 citations (OpenAlex)

Read the paper

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
Apply ISR-Mean post-processing to ERM featuresResNet-50 ERM · Waterbirdsworst-group accuracy: 62.93±5.37 -> 76.10±1.11Improved
Apply ISR-Cov post-processing to ERM featuresResNet-50 ERM · Waterbirdsworst-group accuracy: 62.93±5.37 -> 82.46±0.55Improved
Apply ISR-Mean post-processing to GroupDRO featuresResNet-50 GroupDRO · Waterbirdsworst-group accuracy: 90.79±0.47 -> 90.42±0.61No clear effect
Apply ISR-Cov post-processing to GroupDRO featuresResNet-50 GroupDRO · Waterbirdsworst-group accuracy: 90.79±0.47 -> 91.00±0.45Improved
Apply ISR-Mean post-processing to ERM featuresResNet-50 ERM · CelebAworst-group accuracy: 46.39±2.42 -> 55.39±6.13Improved
Apply ISR-Cov post-processing to ERM featuresResNet-50 ERM · CelebAworst-group accuracy: 46.39±2.42 -> 79.73±5.00Improved
Apply ISR-Mean post-processing to reweighting featuresResNet-50 Reweighting · CelebAworst-group accuracy: 84.44±1.66 -> 90.08±0.50Improved
Apply ISR-Mean post-processing to GroupDRO featuresResNet-50 GroupDRO · CelebAworst-group accuracy: 88.22±1.67 -> 90.95±0.32Improved
Apply ISR-Mean post-processing to ERM featuresBERT ERM · MultiNLIworst-group accuracy: 65.95±1.65 -> 72.60±1.09Improved
Apply ISR-Cov post-processing to ERM featuresBERT ERM · MultiNLIworst-group accuracy: 65.95±1.65 -> 74.21±2.55Improved
Apply ISR-Mean post-processing to GroupDRO featuresBERT GroupDRO · MultiNLIworst-group accuracy: 78.43±0.87 -> 78.95±0.95No clear effect
Apply ISR-Mean to ERM-trained features with only 10% environment labelsWaterbirds ERM · Waterbirdsworst-group accuracy: 73.4%; 10.5%Improved

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