Noise-Adaptive Conformal Classification with Marginal Coverage

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Bortolotti, Teresa, Wang, Y. X. Rachel, Tong, Xin, Menafoglio, Alessandra, Vantini, Simone, Sesia, Matteo
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912210596397056
author Bortolotti, Teresa
Wang, Y. X. Rachel
Tong, Xin
Menafoglio, Alessandra
Vantini, Simone
Sesia, Matteo
author_facet Bortolotti, Teresa
Wang, Y. X. Rachel
Tong, Xin
Menafoglio, Alessandra
Vantini, Simone
Sesia, Matteo
contents Conformal inference provides a rigorous statistical framework for uncertainty quantification in machine learning, enabling well-calibrated prediction sets with precise coverage guarantees for any classification model. However, its reliance on the idealized assumption of perfect data exchangeability limits its effectiveness in the presence of real-world complications, such as low-quality labels -- a widespread issue in modern large-scale data sets. This work tackles this open problem by introducing an adaptive conformal inference method capable of efficiently handling deviations from exchangeability caused by random label noise, leading to informative prediction sets with tight marginal coverage guarantees even in those challenging scenarios. We validate our method through extensive numerical experiments demonstrating its effectiveness on synthetic and real data sets, including CIFAR-10H and BigEarthNet.
format Preprint
id arxiv_https___arxiv_org_abs_2501_18060
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Noise-Adaptive Conformal Classification with Marginal Coverage
Bortolotti, Teresa
Wang, Y. X. Rachel
Tong, Xin
Menafoglio, Alessandra
Vantini, Simone
Sesia, Matteo
Methodology
Machine Learning
Conformal inference provides a rigorous statistical framework for uncertainty quantification in machine learning, enabling well-calibrated prediction sets with precise coverage guarantees for any classification model. However, its reliance on the idealized assumption of perfect data exchangeability limits its effectiveness in the presence of real-world complications, such as low-quality labels -- a widespread issue in modern large-scale data sets. This work tackles this open problem by introducing an adaptive conformal inference method capable of efficiently handling deviations from exchangeability caused by random label noise, leading to informative prediction sets with tight marginal coverage guarantees even in those challenging scenarios. We validate our method through extensive numerical experiments demonstrating its effectiveness on synthetic and real data sets, including CIFAR-10H and BigEarthNet.
title Noise-Adaptive Conformal Classification with Marginal Coverage
topic Methodology
Machine Learning
url https://arxiv.org/abs/2501.18060