Conformal Classification with Equalized Coverage for Adaptively Selected Groups
This addresses uncertainty and fairness issues in classification for applications requiring reliable and equitable predictions, though it appears incremental as it builds on existing conformal inference techniques.
The paper tackles the problem of uncertainty evaluation in classification by developing a conformal inference method that generates prediction sets with valid coverage conditional on adaptively selected features, aiming to balance efficiency and algorithmic fairness. It demonstrates the method's validity and effectiveness on simulated and real datasets.
This paper introduces a conformal inference method to evaluate uncertainty in classification by generating prediction sets with valid coverage conditional on adaptively chosen features. These features are carefully selected to reflect potential model limitations or biases. This can be useful to find a practical compromise between efficiency -- by providing informative predictions -- and algorithmic fairness -- by ensuring equalized coverage for the most sensitive groups. We demonstrate the validity and effectiveness of this method on simulated and real data sets.