AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Group Robust Classification Without Any Group Information

Christos Tsirigotis, Joao Monteiro, Pau Rodriguez, David Vazquez, Aaron Courville

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 4 citations (OpenAlex)

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 uLA instead of the strongest listed baseline, LCcMNIST · cMNIST 0.5% and 5.0%group-balanced accuracy (%): 71.25 -> 75.13; 91.16 -> 92.79Improved
Use uLA instead of LCcMNIST · cMNIST 1.0% and 2.0%group-balanced accuracy (%): 82.25 -> 81.80; 86.21 -> 84.79Hurt
Use uLA instead of the strongest listed baselinecCIFAR10 · cCIFAR10 0.5%, 1.0%, 2.0%, and 5.0%group-balanced accuracy (%): 34.56 -> 34.39; 38.30 -> 62.49; 47.81 -> 63.88; 57.32 -> 74.49Mixed
Use uLA instead of GroupDROWaterbirds · Waterbirdsworst-group accuracy (%): 91.4 -> 86.1Hurt
Use uLA instead of GroupDROCelebA · CelebAworst-group accuracy (%): 88.9 -> 86.5Hurt
Use uLA instead of the strongest listed baseline for each splitsMPI3D · sMPI3D C=2, C=3, C=4, and C=5group-balanced accuracy (%): 31.94 -> 59.58; 48.93 -> 80.53; 71.94 -> 91.11; 94.62 -> 98.05Improved
Replace MoCoV2+ pretraining with BYOL (ablation)uLA · cCIFAR10 1.0%group-balanced accuracy (%): 62.49 -> 59.73No clear effect
Replace MoCoV2+ pretraining with Barlow Twins (ablation)uLA · cCIFAR10 1.0%group-balanced accuracy (%): 62.49 -> 50.08Hurt
Use SSL-pretrained backbone rather than training from scratch (ablation)uLA and the compared training paradigms · CelebA, cCIFAR10, and sMPI3Drobust accuracyImproved
Apply logit adjustment rather than vanilla cross-entropy finetuning (ablation)pretrained backbone · CelebA and cCIFAR10robust accuracyImproved
Use uLA rather than supervised-bias sLA (ablation)logit-adjustment training · CelebA and cCIFAR10robust accuracyNo clear effect
Use uLA rather than supervised-bias sLA (ablation)logit-adjustment training · sMPI3Drobust accuracyMixed

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