Practical Adversarial Multivalid Conformal Prediction
This addresses the problem of reliable uncertainty quantification in sequential prediction for machine learning practitioners, offering a practical and robust alternative to existing methods.
The paper introduces MVP, a lightweight conformal prediction method that achieves adversarial coverage guarantees without a held-out validation set, and provides stronger than marginal coverage with threshold calibration and conditional coverage on user-specified feature subsets.
We give a simple, generic conformal prediction method for sequential prediction that achieves target empirical coverage guarantees against adversarially chosen data. It is computationally lightweight -- comparable to split conformal prediction -- but does not require having a held-out validation set, and so all data can be used for training models from which to derive a conformal score. It gives stronger than marginal coverage guarantees in two ways. First, it gives threshold calibrated prediction sets that have correct empirical coverage even conditional on the threshold used to form the prediction set from the conformal score. Second, the user can specify an arbitrary collection of subsets of the feature space -- possibly intersecting -- and the coverage guarantees also hold conditional on membership in each of these subsets. We call our algorithm MVP, short for MultiValid Prediction. We give both theory and an extensive set of empirical evaluations.