Large-scale probabilistic predictors with and without guarantees of validity

Vladimir Vovk, Ivan Petej, Valentina Fedorova

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper studies theoretically and empirically a method of turning machine learning algorithms into probabilistic predictors that automatically enjoys a property of validity (perfect calibration), is computationally efficient, and preserves predictive efficiency. The price to pay for perfect calibration is that these probabilistic predictors produce imprecise (in practice, almost precise for large data sets) probabilities. When these imprecise probabilities are merged into precise probabilities, the resulting predictors, while losing the theoretical property of perfect calibration, consistently outperform the existing methods in empirical studies.
Original languageEnglish
Title of host publicationNIPS'15
Subtitle of host publicationProceedings of the 28th International Conference on Neural Information Processing Systems
PublisherMIT Press
Pages892-900
Number of pages9
Volume1
DOIs
Publication statusPublished - 7 Dec 2015

Cite this