AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

Provably Invariant Learning without Domain Information

Xiaoyu Tan, LIN Yong, Shengyu Zhu, Chao Qu, Xihe Qiu, Xu Yinghui, Peng Cui, Yuan Qi

ICML 2023 · Label-efficient and robust learning › OOD and anomaly detection · approach: Out-of-distribution generalization theory

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
Feature selection (TIVA-s) instead of feature learning (TIVA-f) on synthetic tabular dataTIVA · Synthetic temporalTest mean accuracy (%)Improved
Environment-independent TIVA (no domain info) vs. ERM on synthetic temporal dataTIVA-s · Synthetic temporalTest mean and worst accuracy: ~15% mean and ~40% worst over ERM (averaged)Improved
Feature learning (TIVA-f) with disentanglement on CelebA pretrained ResNet featuresTIVA · CelebATest worst accuracy (%): 60.32 vs 47.58 for ERMImproved
Feature learning (TIVA-f) on house price dataTIVA · House priceTest mean MSE: 0.305 vs 0.3339 for ZINImproved
Feature selection (TIVA-s) vs feature learning on LandcoverTIVA · LandcoverOOD test accuracy (%): 68.31 vs 68.26No clear effect
Feature learning (TIVA-f) on CelebA: lower train accuracy and worse than ERM on mean test accuracyTIVA · CelebATest mean accuracy (%): 72.60 vs 70.76 ERM (ERM higher on train)Hurt
Feature selection (TIVA-s) on CelebATIVA · CelebATest worst accuracy (%): 45.67 vs 47.58 ERMHurt
Feature selection on house price (TIVA-s) vs feature learningTIVA · House priceTest mean MSE: 0.3276 vs 0.3050Hurt

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