Anytime-valid, Bayes-assisted, Prediction-Powered Inference

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Kilian, Valentin, Cortinovis, Stefano, Caron, François
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911228808396800
author Kilian, Valentin
Cortinovis, Stefano
Caron, François
author_facet Kilian, Valentin
Cortinovis, Stefano
Caron, François
contents Given a large pool of unlabelled data and a smaller amount of labels, prediction-powered inference (PPI) leverages machine learning predictions to increase the statistical efficiency of confidence interval procedures based solely on labelled data, while preserving fixed-time validity. In this paper, we extend the PPI framework to the sequential setting, where labelled and unlabelled datasets grow over time. Exploiting Ville's inequality and the method of mixtures, we propose prediction-powered confidence sequence procedures that are asymptotically valid uniformly over time and naturally accommodate prior knowledge on the quality of the predictions to further boost efficiency. We carefully illustrate the design choices behind our method and demonstrate its effectiveness in real and synthetic examples.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18000
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Anytime-valid, Bayes-assisted, Prediction-Powered Inference
Kilian, Valentin
Cortinovis, Stefano
Caron, François
Machine Learning
Statistics Theory
Given a large pool of unlabelled data and a smaller amount of labels, prediction-powered inference (PPI) leverages machine learning predictions to increase the statistical efficiency of confidence interval procedures based solely on labelled data, while preserving fixed-time validity. In this paper, we extend the PPI framework to the sequential setting, where labelled and unlabelled datasets grow over time. Exploiting Ville's inequality and the method of mixtures, we propose prediction-powered confidence sequence procedures that are asymptotically valid uniformly over time and naturally accommodate prior knowledge on the quality of the predictions to further boost efficiency. We carefully illustrate the design choices behind our method and demonstrate its effectiveness in real and synthetic examples.
title Anytime-valid, Bayes-assisted, Prediction-Powered Inference
topic Machine Learning
Statistics Theory
url https://arxiv.org/abs/2505.18000