Conformal Selective Prediction with General Risk Control

TL;DR

SCoRE is a novel framework that uses conformal inference and e-values to enable strict, finite-sample risk control in selective AI predictions, applicable across various models and settings.

Contribution

It introduces a universal, data-exchangeability-based approach for risk control in selective prediction, extending to distribution shifts without modeling assumptions.

Findings

Effective risk control demonstrated in simulations

Successful application to drug discovery and health prediction

Robustness to distribution shifts shown in experiments

Abstract

In deploying artificial intelligence (AI) models, selective prediction offers the option to abstain from making a prediction when uncertain about model quality. To fulfill its promise, it is crucial to enforce strict and precise error control over cases where the model is trusted. We propose Selective Conformal Risk control with E-values (SCoRE), a new framework for deriving such decisions for any trained model and any user-defined, bounded and continuously-valued risk. SCoRE offers two types of guarantees on the risk among ``positive'' cases in which the system opts to trust the model. Built upon conformal inference and hypothesis testing ideas, SCoRE first constructs a class of (generalized) e-values, which are non-negative random variables whose product with the unknown risk has expectation no greater than one. Such a property is ensured by data exchangeability without requiring any modeling assumptions. Passing these e-values on to hypothesis testing procedures, we yield the binary trust decisions with finite-sample error control. SCoRE avoids the need of uniform concentration, and can be readily extended to settings with distribution shifts. We evaluate the proposed methods with simulations and demonstrate their efficacy through applications to error management in drug discovery, health risk prediction, and large language models.