Aggregating conformal prediction sets via 𝜶-allocation
Authors
Research Topics
Paper Information
-
Journal:
Biometrika -
DOI:
10.1093/biomet/asag048 -
Published:
July 21, 2026 -
Added to Tracker:
Jul 22, 2026
Abstract
Abstract Conformal prediction offers a distribution-free framework for constructing prediction sets with finite-sample coverage. Yet, efficiently leveraging multiple nonconformity scores to reduce set sizes remains an open challenge. Instead of selecting a single best score, this work introduces a principled aggregation strategy that intersects multiple conformal prediction sets whose confidence levels are optimally allocated to minimize empirical set size while maintaining asymptotic coverage. Two variants are developed to guarantee finite-sample coverage via sample splitting and full conformalization, respectively. An individualized allocation strategy is further proposed to promote local size efficiency while achieving asymptotic conditional coverage. Extensive experiments on synthetic and real-world datasets demonstrate that our approach achieves considerably smaller prediction sets than state-of-the-art baselines while maintaining valid coverage.
Author Details
Changliang Zou
AuthorZhaojun Wang
AuthorCongbin Xu
AuthorYue Yu
AuthorHaojie Ren
AuthorResearch Topics & Keywords
Statistical Learning
Research AreaCitation Information
APA Format
Changliang Zou
,
Zhaojun Wang
,
Congbin Xu
,
Yue Yu
&
Haojie Ren
(2026)
.
Aggregating conformal prediction sets via 𝜶-allocation.
Biometrika
, 10.1093/biomet/asag048.
BibTeX Format
@article{paper1479,
title = { Aggregating conformal prediction sets via 𝜶-allocation },
author = {
Changliang Zou
and Zhaojun Wang
and Congbin Xu
and Yue Yu
and Haojie Ren
},
journal = { Biometrika },
year = { 2026 },
doi = { 10.1093/biomet/asag048 },
url = { https://doi.org/10.1093/biomet/asag048 }
}