Spuriosity Didn't Kill the Classifier: Using Invariant Predictions to Harness Spurious Features
10 December 2023 · 1 min read
Instead of discarding “spurious” features whose relationship with the label changes across domains, we show they can be safely exploited in the test domain without labels: pseudo-labels from stable features provide sufficient guidance when stable and unstable features are conditionally independent given the label. Our Stable Feature Boosting algorithm learns an asymptotically-optimal predictor without test-domain labels.