Shape-Adaptive Conditional Calibration for Conformal Prediction via Minimax Optimization

Yajie Bao, Chuchen Zhang, Zhaojun Wang, Haojie Ren, Changliang Zou · 2026

Read Shape-Adaptive Conditional Calibration for Conformal Prediction via Minimax Optimization by Yajie Bao, Chuchen Zhang, Zhaojun Wang, Haojie Ren, Changliang Zou (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-2603-23374.txt.

Introduction

Bao and four co-authors tackle conditional coverage in conformal prediction. A prediction set should cover the truth at the promised rate for each case, not just on average. Past tools calibrated one fixed set shape, which limits how sets can bend. Their framework, MOPI, optimizes over a flexible family of set shapes at calibration time. A minimax setup keeps it tied to low squared coverage error. Theory gives oracle bounds and an optimal error rate. Tests on tricky distributions show tighter sets than rival methods.

Catalog record

CREATORYajie Bao, Chuchen Zhang, Zhaojun Wang, Haojie Ren, Changliang Zou
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE207111
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.23374

Browse the full library · About the Tower of Babel