Robust Conditional Conformal Prediction via Branched Normalizing Flow

Rui Xu, Xingyuan Chen, Wenxing Huang, Minxuan Huang, Weiyan Chen, Sihong Xie, Hui Xiong · 2026

Read Robust Conditional Conformal Prediction via Branched Normalizing Flow by Rui Xu, Xingyuan Chen, Wenxing Huang, Minxuan Huang, Weiyan Chen, Sihong Xie, Hui Xiong (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-2605-01868.txt.

Description

Rui Xu and colleagues work on conformal prediction, a method that gives prediction sets with a stated coverage rate. The guarantee assumes test data match the calibration data. When the distribution shifts, most fixes align overall score patterns, which can still leave single inputs badly covered in places where conditions differ.

The authors bound that single-input error by the distance between calibration and test distributions, which points to transport as a fix. Their Branched Normalizing Flow has two branches: one maps a test input toward the calibration form, and the other maps its prediction set back. Tests show the method holds conditional coverage better under shift than prior approaches.

Catalog record

CREATORRui Xu, Xingyuan Chen, Wenxing Huang, Minxuan Huang, Weiyan Chen, Sihong Xie, Hui Xiong
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE104 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.01868

Browse the full library · About the Tower of Babel