@inproceedings{khodja-etal-2026-anomaly,
title = "Anomaly Detection using Source-Reversed Flow Matching",
author = {Khodja, Amine Youcef and
Pantin, Jeremie and
Dias, Ga{\"e}l and
Maurel, Fabrice},
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.2322/",
pages = "52746--52767",
ISBN = "979-8-89176-459-0",
abstract = "Anomaly detection in textual data remains an underexplored problem, hindered by fragmented benchmarks, scarce labeled data, and methods that fail to model the distribution of normal text in a continuous and generative manner. We propose Flocat, an unsupervised approach that reformulates this task through a source-reversed flow matching paradigm: rather than learning to generate, we learn a flow from inlier text embeddings toward a compact isotropic Gaussian target, whose radius is explicitly constrained to ensure that the distance to the centroid serves as a geometrically consistent anomaly score. Built on a Diffusion Transformer (DiT) backbone, Flocat operates at both token and sentence level through a level-agnostic interpolation mechanism, enabling fine-grained signal interpretability. We evaluate on 10 datasets spanning text classification, spam detection, sentiment analysis, human- vs. machine-generated text and depression screening, consistently achieving state-of-the-art AUC-ROC across all benchmarks and text representations."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="khodja-etal-2026-anomaly">
<titleInfo>
<title>Anomaly Detection using Source-Reversed Flow Matching</title>
</titleInfo>
<name type="personal">
<namePart type="given">Amine</namePart>
<namePart type="given">Youcef</namePart>
<namePart type="family">Khodja</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jeremie</namePart>
<namePart type="family">Pantin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gaël</namePart>
<namePart type="family">Dias</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fabrice</namePart>
<namePart type="family">Maurel</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Anomaly detection in textual data remains an underexplored problem, hindered by fragmented benchmarks, scarce labeled data, and methods that fail to model the distribution of normal text in a continuous and generative manner. We propose Flocat, an unsupervised approach that reformulates this task through a source-reversed flow matching paradigm: rather than learning to generate, we learn a flow from inlier text embeddings toward a compact isotropic Gaussian target, whose radius is explicitly constrained to ensure that the distance to the centroid serves as a geometrically consistent anomaly score. Built on a Diffusion Transformer (DiT) backbone, Flocat operates at both token and sentence level through a level-agnostic interpolation mechanism, enabling fine-grained signal interpretability. We evaluate on 10 datasets spanning text classification, spam detection, sentiment analysis, human- vs. machine-generated text and depression screening, consistently achieving state-of-the-art AUC-ROC across all benchmarks and text representations.</abstract>
<identifier type="citekey">khodja-etal-2026-anomaly</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2322/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>52746</start>
<end>52767</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Anomaly Detection using Source-Reversed Flow Matching
%A Khodja, Amine Youcef
%A Pantin, Jeremie
%A Dias, Gaël
%A Maurel, Fabrice
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F khodja-etal-2026-anomaly
%X Anomaly detection in textual data remains an underexplored problem, hindered by fragmented benchmarks, scarce labeled data, and methods that fail to model the distribution of normal text in a continuous and generative manner. We propose Flocat, an unsupervised approach that reformulates this task through a source-reversed flow matching paradigm: rather than learning to generate, we learn a flow from inlier text embeddings toward a compact isotropic Gaussian target, whose radius is explicitly constrained to ensure that the distance to the centroid serves as a geometrically consistent anomaly score. Built on a Diffusion Transformer (DiT) backbone, Flocat operates at both token and sentence level through a level-agnostic interpolation mechanism, enabling fine-grained signal interpretability. We evaluate on 10 datasets spanning text classification, spam detection, sentiment analysis, human- vs. machine-generated text and depression screening, consistently achieving state-of-the-art AUC-ROC across all benchmarks and text representations.
%U https://aclanthology.org/2026.emnlp-main.2322/
%P 52746-52767
Markdown (Informal)
[Anomaly Detection using Source-Reversed Flow Matching](https://aclanthology.org/2026.emnlp-main.2322/) (Khodja et al., EMNLP 2026)
ACL
- Amine Youcef Khodja, Jeremie Pantin, Gaël Dias, and Fabrice Maurel. 2026. Anomaly Detection using Source-Reversed Flow Matching. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 52746–52767, Budapest, Hungary. Association for Computational Linguistics.