A Post-Processing Conformal Prediction Approach for Conditional Coverage via Pivotal Scores

Félix Laplante[0.5em] Université Paris-Saclay, CNRS, Univ Évry,Laboratoire de Mathématiques et Modélisation d’Évry,91037, Évry-Courcouronnes, France. [email protected] · 2026

Read A Post-Processing Conformal Prediction Approach for Conditional Coverage via Pivotal Scores by Félix Laplante[0.5em] Université Paris-Saclay, CNRS, Univ Évry,Laboratoire de Mathématiques et Modélisation d’Évry,91037, Évry-Courcouronnes, France. [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-25852.txt.

Description

Laplante tackles conditional coverage: prediction sets should cover every subgroup, not just on average. Exact conditional validity needs assumptions. Yet for data drawn independently and identically it comes down to a feature-free score. His PIT-CP turns any base score into a nearly invariant one, keeping its shape and coverage. The task then becomes 1D density estimation on the score, rather than full estimation. On varied datasets PIT-CP matches or beats strategies.

Catalog record

CREATORFélix Laplante[0.5em] Université Paris-Saclay, CNRS, Univ Évry,Laboratoire de Mathématiques et Modélisation d’Évry,91037, Évry-Courcouronnes, France. [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE99689
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2605.25852

Browse the full library · About the Tower of Babel