CoChat

A Survey on Explainable Anomaly Detection

Open access
Li, Zhong, Zhu, Yuxuan, Van Leeuwen, Matthijs · 2024 · ACM Transactions on Knowledge Discovery from Data

The DOI matches its Crossref record; 228 of the 228 references it cites were found in a registry or an index; 45 of the 131 sources its claims cite were checked for support.

Abstract

This survey provides a comprehensive and structured survey of state-of-the-art explainable anomaly detection techniques.

Overview

CoChat Summary

TL;DR

This paper is a survey of explainable anomaly detection (XAD) that reviews more than 150 articles and proposes a six-criteria taxonomy to organize the field. It argues that XAD is needed because anomaly detection is widely used in high-stakes settings, but explanations must be faithful, comprehensible, scalable, and aligned with the actual anomaly definition.

Key findings

  • The survey reviews more than 150 articles on XAD and organizes them into four high-level categories and twelve fine-grained categories.
  • It proposes six main taxonomy criteria for XAD methods: explanation type, model-agnostic vs. model-specific, data type, local vs. global explanation, techniques used, and applicable data type.
  • The authors report that pre-model techniques are generally model-agnostic and provide global explanations, while in-model techniques are model-specific and often provide feature-based explanations.
  • They note that post-model techniques are often local and can suffer from poor explanation fidelity, especially when the explanation definition differs from the detection definition.

Methods

  • The paper is a comparative, iterative literature survey with three cycles of collection and screening.
  • It searches Google Scholar, IEEE Xplore, ACM Digital Library, DBLP, and Web of Science, covering publications from January 1998 to February 2022.
  • The survey classifies methods using a taxonomy built from six criteria and then reviews the literature in four sections: pre-model, in-model, shallow post-model, and deep post-model techniques.

Limitations

  • The paper states there is no consensus on what a good XAD technique should be, especially for evaluating explanation quality.
  • It states that evaluation of XAD remains unresolved, and that methods developed for classification or clustering are non-trivial to extend to anomaly detection.
  • Evident from the text: because the survey is based on literature up to February 2022 and explicitly says it does not claim to have covered all relevant publications, it may not include later work or every relevant method.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3609333

Cites

228 of 228 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

20 Valid · 48 Not valid · 74 Not checked · 116 Own results and interpretations

118 more claims · 16 with their cited sources checked

Create a free account to see all 258 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 108
  • It may state two separate claims as one. 34
  • The quote only partly states this claim. 19

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

3 of 3

Citations

Cites

Show all 228 references

Research works printed but not linked: 22

  1. 10.5555/645920.672836 · doi:10.5555/645920.672836
  2. Xingyan Bin Ying Zhao and Bilong Shen. 2016. Abnormal subspace sparse PCA for anomaly detection and interpretation. In ODDx3’15:ACM SIGKDD 2015 Workshop . Association for Computing Machinery.
  3. Chandranil Chakraborttii and Heiner Litz. 2020. Explaining SSD failures using anomaly detection. In Proceedings of the Non-Volatile Memory Workshop. 1.
  4. N. F. Chen, Zhiyuan Du, and Khin Hua Ng. 2018. Scene graphs for interpretable video anomaly classification. In Proceedings of the Conference on Neural Information Processing Systems Workshop on Visually Grounded Interaction and Language.
  5. Ian Davidson. 2007. Anomaly detection, explanation and visualization. SGI, Tokyo, Japan, Tech. Rep (2007).
  6. Levent Ertoz, Eric Eilertson, Aleksandar Lazarevic, Pang-Ning Tan, Vipin Kumar, Jaideep Srivastava, and Paul Dokas. 2004. Minds-minnesota intrusion detection system. Next Generation Data Mining (2004), 199–218.
  7. Nico Gnoss Martin Schultz and Marina Tropmann-Frick. 2022. XAI in the audit domain-explaining an autoencoder model for anomaly detection. In 17th International Conference on Wirtschaftsinformatik . AISeL.
  8. Nirmal Sobha Kartha Clément Gautrais and Vincent Vercruyssen. 2021. Why are you weird? Infusing interpretability in isolation forest for anomaly detection. In Proceedings of the Explainable Agency in AI Workshop (AAAI’21) . 51–57.
  9. Guolin Ke Qi Meng Thomas Finley Taifeng Wang Wei Chen Weidong Ma Qiwei Ye and Tie-Yan Liu. 2017. Lightgbm: A highly efficient gradient boosting decision tree. In Proceedings of the 31st International Conference on Neural Information Processing Systems (Long Beach California USA) (NIPS’17) . Curran Associates Inc. 3149–3157.
  10. Edwin M. Knorr and Raymond T. Ng. 1998. Algorithms for mining distance-based outliers in large datasets. In Proceedings of the VLDB. Citeseer, 392–403.
  11. Edwin M. Knorr and Raymond T. Ng. 1999. Finding intensional knowledge of distance-based outliers. In Proceedings of the VLDB. Citeseer, 211–222.
  12. Martin Kopp, Tomáš Pevnỳ, and Martin Holena. 2014. Interpreting and clustering outliers with sapling random forests. In Proceedings of the European Conference on Information Technologies–Applications and Theory.61–67.
  13. Yinxi Liu Kai Yang Shaoyu Dou and Pan Luo. 2022. Interpretable anomaly detection in variable-length co-evolving rhythmic sequences. In 1st Workshop on Artificial Intelligence for Anomalies and Novelties (AI4AN) .
  14. Scott M. Lundberg and Su-In Lee. 2017. A unified approach to interpreting model predictions. In Proceedings of the 31st International Conference on Neural Information Processing Systems (Long Beach California USA) (NIPS’17) . Curran Associates Inc. Red Hook NY USA 4768–4777.
  15. Alireza Makhzani Jonathon Shlens Navdeep Jaitly Ian Goodfellow and Brendan Frey. 2016. Adversarial autoencoders. ICLR Workshop track .
  16. Guansong Pang, Longbing Cao, and Ling Chen. 2016. Outlier detection in complex categorical data by modelling the feature value couplings. In Proceedings of the IJCAI International Joint Conference on Artificial Intelligence.
  17. Tomáš Pevnỳ and Martin Kopp. 2014. Explaining anomalies with sapling random forests. In Proceedings of the Information Technologies-Applications and Theory Workshops, Posters, and Tutorials (ITAT’14).
  18. Daniel M. Roy and Yee Whye Teh. 2008. The mondrian process. In Proceedings of the NIPS, Vol. 21.
  19. Bernhard Schölkopf Robert C. Williamson Alex Smola John Shawe-Taylor and John Platt. 1999. Support vector method for novelty detection. In Proceedings of the 12th International Conference on Neural Information Processing Systems (Denver CO) (NIPS’99 Vol. 12) . MIT Press Cambridge MA USA 582–588.
  20. John Sipple. 2020. Interpretable, multidimensional, multimodal anomaly detection with negative sampling for detection of device failure. In Proceedings of the International Conference on Machine Learning. PMLR, 9016–9025.
  21. Mukund Sundararajan and Amir Najmi. 2020. The many Shapley values for model explanation. In Proceedings of the International Conference on Machine Learning. PMLR, 9269–9278.
  22. Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In Proceedings of the International Conference on Machine Learning. PMLR, 3319–3328.

Cited by

Metadata

Authors
Li, Z., Zhu, Y., & Van Leeuwen, M.
Affiliations
  • Leiden University (NL)
  • Leiden Institute of Advanced Computer Science (LIACS), Leiden University, The Netherlands
Venue
ACM Transactions on Knowledge Discovery from Data
Published
2023-09-06
Volume
18
Issue
1
Pages
1-54
ISSN
  • 1556-4681 (ISSN-L)
  • 1556-472X
OpenAlex
W4384407578
Cited by
139
References
203
Citations by year
2026: 47 · 2025: 50 · 2024: 36 · 2023: 6
FWCI
13.07
Type
journal article
Language
en
Open access
Locations
  • ACM Transactions on Knowledge Discovery from Data (publishedVersion, cc-by-sa)
  • arXiv (Cornell University) (submittedVersion)
  • Leiden Repository (Leiden University) (submittedVersion)
  • arXiv (Cornell University)
Funders
Dutch Research Council
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
  • Physical Sciences › Computer Science › Artificial Intelligence › Anomaly Detection Techniques and Applications
  • Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
Keywords
explainable anomaly detection, anomaly detection, explainability, safety-critical domains

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-08, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-08 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.