AI Paper Atlas

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

Spuriosity Didn’t Kill the Classifier: Using Invariant Predictions to Harness Spurious Features

Cian Eastwood, Shashank Singh, Andrei Liviu Nicolicioiu, Marin Vlastelica, Julius von Kügelgen, Bernhard Schölkopf

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 2 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
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · Synthetic ACaccuracy: 74.7 ± 1.2 -> 89.2 ± 2.9Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · Synthetic CE-DDaccuracy: 74.9 ± 3.6 -> 88.6 ± 1.4Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · CMNISTaccuracy: 70.6 ± 1.8 -> 88.1 ± 1.8Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · PACS Paccuracy: 93.7 ± 0.6 -> 95.8 ± 0.6Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · PACS Aaccuracy: 78.1 ± 1.1 -> 80.4 ± 1.3Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · PACS Caccuracy: 73.7 ± 0.6 -> 76.6 ± 0.6Improved
adapt the unstable-feature predictor using unlabeled test-domain data (ablation)SFB · PACS Saccuracy: 69.7 ± 2.3 -> 71.8 ± 2.0Improved
add bias correction to test-domain adaptation (ablation)SFB · CMNIST test domains with varying color-label correlationaccuracyImproved
add post-hoc calibration of the stable predictor (ablation)SFB · CMNIST test domains with varying color-label correlationaccuracyImproved
use multiple pseudo-labeling rounds (ablation)SFB · CMNIST test domains with varying color-label correlationaccuracyImproved
use Stable Feature Boosting instead of ERMdomain generalization · Synthetic ACaccuracy: 9.9 ± 0.1 -> 89.2 ± 2.9Improved
use Stable Feature Boosting instead of ERMdomain generalization · Synthetic CE-DDaccuracy: 11.6 ± 0.7 -> 88.6 ± 1.4Improved

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