Saved in:
Bibliographic Details
Main Authors: Lee, Junu, Popov, Ilia, Ren, Zhimei
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.02703
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910144275677184
author Lee, Junu
Popov, Ilia
Ren, Zhimei
author_facet Lee, Junu
Popov, Ilia
Ren, Zhimei
contents This paper presents a powerful methodology for flexible full-data nonparametric novelty detection that offers distribution-free false discovery rate (FDR) control guarantees. Building on the full conformal inference framework and the concept of e-values, we introduce full conformal e-values to quantify evidence for novelty relative to a given reference dataset. These e-values are then utilized by carefully crafted multiple testing procedures to identify a set of novel units out-of-sample with provable finite-sample FDR control. We showcase several instantiations of e-values, including those which employ a data-driven model selection strategy to amplify power. Furthermore, our framework is extended to address distribution shift, accommodating scenarios where novelty detection must be performed on data drawn from a shifted distribution relative to the reference dataset. In all settings, our method can perform powerfully -- outperforming existing novelty detection methods -- even with limited amounts of reference data; this is illustrated by empirical evaluations on synthetic data and an application to a malicious LLM prompts dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2501_02703
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Full-conformal novelty detection
Lee, Junu
Popov, Ilia
Ren, Zhimei
Methodology
This paper presents a powerful methodology for flexible full-data nonparametric novelty detection that offers distribution-free false discovery rate (FDR) control guarantees. Building on the full conformal inference framework and the concept of e-values, we introduce full conformal e-values to quantify evidence for novelty relative to a given reference dataset. These e-values are then utilized by carefully crafted multiple testing procedures to identify a set of novel units out-of-sample with provable finite-sample FDR control. We showcase several instantiations of e-values, including those which employ a data-driven model selection strategy to amplify power. Furthermore, our framework is extended to address distribution shift, accommodating scenarios where novelty detection must be performed on data drawn from a shifted distribution relative to the reference dataset. In all settings, our method can perform powerfully -- outperforming existing novelty detection methods -- even with limited amounts of reference data; this is illustrated by empirical evaluations on synthetic data and an application to a malicious LLM prompts dataset.
title Full-conformal novelty detection
topic Methodology
url https://arxiv.org/abs/2501.02703