Adaptive novelty detection with false discovery rate guarantee
The Annals of Statistics · 01.02.2024 · 52 (1) · 157-183
DOI: 10.1214/23-aos2338 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Annals of Statistics, erschienen am 01.02.2024. Verfasst von Ariane Marandon und 3 weiteren Autor:innen. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 72 Literaturangaben.
Abstract (Originalsprache)
This paper studies the semisupervised novelty detection problem where a set of “typical” measurements is available to the researcher. Motivated by recent advances in multiple testing and conformal inference, we propose AdaDetect, a flexible method that is able to wrap around any probabilistic classification algorithm and control the false discovery rate (FDR) on detected novelties in finite samples without any distributional assumption other than exchangeability. In contrast to classical FDR-controlling procedures that are often committed to a pre-specified p-value function, AdaDetect learns the transformation in a data-adaptive manner to focus the power on the directions that distinguish between inliers and outliers. Inspired by the multiple testing literature, we further propose variants of AdaDetect that are adaptive to the proportion of nulls while maintaining the finite-sample FDR control. The methods are illustrated on synthetic datasets and real-world datasets, including an application in astrophysics.
Zitieren
APA
Marandon, A., Lei, L., Mary, D., & Roquain, E. (2024). Adaptive novelty detection with false discovery rate guarantee. The Annals of Statistics, 52(1), 157-183. https://doi.org/10.1214/23-aos2338
Vancouver
Marandon A, Lei L, Mary D, Roquain E. Adaptive novelty detection with false discovery rate guarantee. The Annals of Statistics. 2024;52(1):157-183. doi:10.1214/23-aos2338
BibTeX
@article{Marandon2024,
author = {Marandon, Ariane and Lei, Lihua and Mary, David and Roquain, Etienne},
title = {Adaptive novelty detection with false discovery rate guarantee},
journal = {The Annals of Statistics},
year = {2024},
volume = {52},
number = {1},
pages = {157--183},
doi = {10.1214/23-aos2338}
}
EUVIMED