@inproceedings{fihey-etal-2026-geometry,
title = "A Geometry-Aware Depth Function for Robust Unsupervised Textual Anomaly Detection",
author = "Fihey, Pierre and
Labeau, Matthieu and
Mozharovskyi, Pavlo",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1731/",
pages = "37011--37030",
ISBN = "979-8-89176-462-0",
abstract = "Textual anomaly detection (AD) aims to identify samples that deviate from the underlying distribution of normal data, without requiring explicit supervision on anomalies. Existing literature has focused almost exclusively on the semi-supervised setting, assuming access to clean training datasets composed solely of normal samples. However, this assumption rarely holds in real-world scenarios, as datasets used to infer the normal distribution are frequently contaminated by anomalies. In this work, we first highlight the sensitivity of conventional anomaly detection methods to contamination, before demonstrating the relevance of \textit{data depth functions} for robust textual anomaly detection. Building on recent findings regarding the geometry of language model embeddings, further supported by our empirical observations, we introduce the \textit{geodesic depth} to better capture the structure of contextual embedding spaces. Our results show that the geodesic depth consistently outperforms commonly used anomaly detection methods once contamination exceeds 10{\%} in the training data, with substantially larger improvements observed in the most challenging scenarios."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fihey-etal-2026-geometry">
<titleInfo>
<title>A Geometry-Aware Depth Function for Robust Unsupervised Textual Anomaly Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pierre</namePart>
<namePart type="family">Fihey</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Matthieu</namePart>
<namePart type="family">Labeau</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pavlo</namePart>
<namePart type="family">Mozharovskyi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Textual anomaly detection (AD) aims to identify samples that deviate from the underlying distribution of normal data, without requiring explicit supervision on anomalies. Existing literature has focused almost exclusively on the semi-supervised setting, assuming access to clean training datasets composed solely of normal samples. However, this assumption rarely holds in real-world scenarios, as datasets used to infer the normal distribution are frequently contaminated by anomalies. In this work, we first highlight the sensitivity of conventional anomaly detection methods to contamination, before demonstrating the relevance of data depth functions for robust textual anomaly detection. Building on recent findings regarding the geometry of language model embeddings, further supported by our empirical observations, we introduce the geodesic depth to better capture the structure of contextual embedding spaces. Our results show that the geodesic depth consistently outperforms commonly used anomaly detection methods once contamination exceeds 10% in the training data, with substantially larger improvements observed in the most challenging scenarios.</abstract>
<identifier type="citekey">fihey-etal-2026-geometry</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1731/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>37011</start>
<end>37030</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Geometry-Aware Depth Function for Robust Unsupervised Textual Anomaly Detection
%A Fihey, Pierre
%A Labeau, Matthieu
%A Mozharovskyi, Pavlo
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F fihey-etal-2026-geometry
%X Textual anomaly detection (AD) aims to identify samples that deviate from the underlying distribution of normal data, without requiring explicit supervision on anomalies. Existing literature has focused almost exclusively on the semi-supervised setting, assuming access to clean training datasets composed solely of normal samples. However, this assumption rarely holds in real-world scenarios, as datasets used to infer the normal distribution are frequently contaminated by anomalies. In this work, we first highlight the sensitivity of conventional anomaly detection methods to contamination, before demonstrating the relevance of data depth functions for robust textual anomaly detection. Building on recent findings regarding the geometry of language model embeddings, further supported by our empirical observations, we introduce the geodesic depth to better capture the structure of contextual embedding spaces. Our results show that the geodesic depth consistently outperforms commonly used anomaly detection methods once contamination exceeds 10% in the training data, with substantially larger improvements observed in the most challenging scenarios.
%U https://aclanthology.org/2026.findings-emnlp.1731/
%P 37011-37030
Markdown (Informal)
[A Geometry-Aware Depth Function for Robust Unsupervised Textual Anomaly Detection](https://aclanthology.org/2026.findings-emnlp.1731/) (Fihey et al., Findings 2026)
ACL