Conformal prediction without knowledge of labeled calibration data

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Flechsig, Jonas, Pilz, Maximilian
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911151426633728
author Flechsig, Jonas
Pilz, Maximilian
author_facet Flechsig, Jonas
Pilz, Maximilian
contents We extend the method of conformal prediction beyond the case relying on labeled calibration data. Replacing the calibration scores by suitable estimates, we identify conformity sets $C$ for classification and regression models that rely on unlabeled calibration data. Given a classification model with accuracy $1-β$, we prove that the conformity sets guarantee a coverage of $P(Y \in C) \geq 1-α-β$ for an arbitrary parameter $α\in (0,1)$. The same coverage guarantee also holds for regression models, if we replace the accuracy by a similar exactness measure. Finally, we describe how to use the theoretical results in practice.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10321
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Conformal prediction without knowledge of labeled calibration data
Flechsig, Jonas
Pilz, Maximilian
Methodology
We extend the method of conformal prediction beyond the case relying on labeled calibration data. Replacing the calibration scores by suitable estimates, we identify conformity sets $C$ for classification and regression models that rely on unlabeled calibration data. Given a classification model with accuracy $1-β$, we prove that the conformity sets guarantee a coverage of $P(Y \in C) \geq 1-α-β$ for an arbitrary parameter $α\in (0,1)$. The same coverage guarantee also holds for regression models, if we replace the accuracy by a similar exactness measure. Finally, we describe how to use the theoretical results in practice.
title Conformal prediction without knowledge of labeled calibration data
topic Methodology
url https://arxiv.org/abs/2509.10321