Enhancing Conformal Prediction Using E-Test Statistics

TL;DR

This paper explores the use of e-test statistics to improve conformal prediction methods, aiming to produce more accurate and reliable prediction intervals in machine learning models.

Contribution

It introduces a novel approach that leverages e-test statistics within conformal prediction, providing an alternative to p-value-based methods.

Findings

Enhanced prediction intervals with better coverage.

Improved efficiency over traditional conformal methods.

Validation on benchmark datasets shows superior performance.

Abstract

Conformal Prediction (CP) serves as a robust framework that quantifies uncertainty in predictions made by Machine Learning (ML) models. Unlike traditional point predictors, CP generates statistically valid prediction regions, also known as prediction intervals, based on the assumption of data exchangeability. Typically, the construction of conformal predictions hinges on p-values. This paper, however, ventures down an alternative path, harnessing the power of e-test statistics to augment the efficacy of conformal predictions by introducing a BB-predictor (bounded from the below predictor).