Conformal Prediction Adaptive to Unknown Subpopulation Shifts
This addresses uncertainty quantification for black-box models in real-world scenarios with hidden subpopulation shifts, offering a practical solution for machine learning applications.
The paper tackles the problem of conformal prediction failing under unknown subpopulation shifts, where test data has a different mix of subpopulations than calibration data without group labels, and proposes new methods that provably adapt to ensure valid coverage, with experiments showing reliable coverage maintenance on vision and language benchmarks.
Conformal prediction is widely used to equip black-box machine learning models with uncertainty quantification, offering formal coverage guarantees under exchangeable data. However, these guarantees fail when faced with subpopulation shifts, where the test environment contains a different mix of subpopulations than the calibration data. In this work, we focus on unknown subpopulation shifts where we are not given group-information i.e. the subpopulation labels of datapoints have to be inferred. We propose new methods that provably adapt conformal prediction to such shifts, ensuring valid coverage without explicit knowledge of subpopulation structure. While existing methods in similar setups assume perfect subpopulation labels, our framework explicitly relaxes this requirement and characterizes conditions where formal coverage guarantees remain feasible. Further, our algorithms scale to high-dimensional settings and remain practical in realistic machine learning tasks. Extensive experiments on vision (with vision transformers) and language (with large language models) benchmarks demonstrate that our methods reliably maintain coverage and effectively control risks in scenarios where standard conformal prediction fails.