Target Strangeness: A Novel Conformal Prediction Difficulty Estimator

TL;DR

This paper presents Target Strangeness, a new difficulty estimator for conformal prediction that improves the normalization of prediction intervals by measuring how atypical a prediction is relative to its neighbors.

Contribution

It introduces Target Strangeness as a novel approach for estimating difficulty in conformal prediction, outperforming existing methods in regression tasks.

Findings

Target Strangeness surpasses state-of-the-art performance in conformal regression.

It effectively normalizes prediction intervals based on local target distribution.

The method demonstrates robustness across multiple experiments.

Abstract

This paper introduces Target Strangeness, a novel difficulty estimator for conformal prediction (CP) that offers an alternative approach for normalizing prediction intervals (PIs). By assessing how atypical a prediction is within the context of its nearest neighbours' target distribution, Target Strangeness can surpass the current state-of-the-art performance. This novel difficulty estimator is evaluated against others in the context of several conformal regression experiments.