Computationally efficient versions of conformal predictive distributions

TL;DR

This paper introduces two computationally efficient variants of conformal predictive systems for regression, enhancing practicality by balancing validity and predictive efficiency.

Contribution

The paper proposes split and cross-conformal predictive systems, improving computational efficiency while maintaining validity or increasing predictive efficiency.

Findings

Split conformal predictive systems guarantee validity.

Cross-conformal predictive systems offer higher predictive efficiency.

Cross-conformal validity is empirical and depends on randomization.

Abstract

Conformal predictive systems are a recent modification of conformal predictors that output, in regression problems, probability distributions for labels of test observations rather than set predictions. The extra information provided by conformal predictive systems may be useful, e.g., in decision making problems. Conformal predictive systems inherit the relative computational inefficiency of conformal predictors. In this paper we discuss two computationally efficient versions of conformal predictive systems, which we call split conformal predictive systems and cross-conformal predictive systems. The main advantage of split conformal predictive systems is their guaranteed validity, whereas for cross-conformal predictive systems validity only holds empirically and in the absence of excessive randomization. The main advantage of cross-conformal predictive systems is their greater predictive efficiency.