# Towards Explaining Anomalies: A Deep Taylor Decomposition of One-Class Models

arXiv: https://arxiv.org/abs/1805.06230 · HTML: https://mlsanity.com/paper/1805.06230/

## Key facts

- Cited by 24 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 May 2018.

## Built on by (most cited first, 24 of 24)

- [This looks more like that: Enhancing Self-Explaining Models by Prototypical Relevance Propagation](https://mlsanity.com/paper/2108.12204/) (2021)
- [A Survey on Explainable Anomaly Detection](https://mlsanity.com/paper/2210.06959/) (2022)
- [Toward Explainable AI for Regression Models](https://mlsanity.com/paper/2112.11407/) (2021)
- [Explainable Intrusion Detection Systems (X-IDS): A Survey of Current Methods, Challenges, and Opportunities](https://mlsanity.com/paper/2207.06236/) (2022)
- [The Clever Hans Effect in Unsupervised Learning](https://mlsanity.com/paper/2408.08041/) (2024)
- [DeepAID: Interpreting and Improving Deep Learning-based Anomaly Detection in Security Applications](https://mlsanity.com/paper/2109.11495/) (2021)
- [Data-Efficient and Interpretable Tabular Anomaly Detection](https://mlsanity.com/paper/2203.02034/) (2022)
- [DI-AA: An Interpretable White-box Attack for Fooling Deep Neural Networks](https://mlsanity.com/paper/2110.07305/) (2021)
- [AI-based Anomaly Detection for Clinical-Grade Histopathological Diagnostics](https://mlsanity.com/paper/2406.14866/) (2024)
- [A general-purpose method for applying Explainable AI for Anomaly Detection](https://mlsanity.com/paper/2207.11564/) (2022)
- [On Root Cause Localization and Anomaly Mitigation through Causal Inference](https://mlsanity.com/paper/2212.04031/) (2022)
- [Trustworthy Anomaly Detection: A Survey](https://mlsanity.com/paper/2202.07787/) (2022)
- [Framing Algorithmic Recourse for Anomaly Detection](https://mlsanity.com/paper/2206.14384/) (2022)
- [Explaining Anomalies using Denoising Autoencoders for Financial Tabular Data](https://mlsanity.com/paper/2209.10658/) (2022)
- [Self-Supervised Training with Autoencoders for Visual Anomaly Detection](https://mlsanity.com/paper/2206.11723/) (2022)
- [Insightful analysis of historical sources at scales beyond human capabilities using unsupervised Machine Learning and XAI](https://mlsanity.com/paper/2310.09091/) (2023)
- [Interpretable Maximum Margin Deep Anomaly Detection](https://mlsanity.com/paper/2603.07073/) (2026)
- [Towards Interpretable Ensemble Learning for Image-based Malware Detection](https://mlsanity.com/paper/2101.04889/) (2021)
- [Towards interpreting ML-based automated malware detection models: a survey](https://mlsanity.com/paper/2101.06232/) (2021)
- [Detection of Dataset Shifts in Learning-Enabled Cyber-Physical Systems using Variational Autoencoder for Regression](https://mlsanity.com/paper/2104.06613/) (2021)
- [A Novel Dual-pooling Attention Module for UAV Vehicle Re-identification](https://mlsanity.com/paper/2306.14104/) (2023)
- [A Safe Screening Rule with Bi-level Optimization of nu Support Vector Machine](https://mlsanity.com/paper/2403.01769/) (2024)
- [Fast and Accurate Explanations of Distance-Based Classifiers by Uncovering Latent Explanatory Structures](https://mlsanity.com/paper/2508.03913/) (2025)
- [Hidden Monotonicity: Explaining Deep Neural Networks via their DC Decomposition](https://mlsanity.com/paper/2601.07700/) (2026)

## Abstract (arXiv)

A common machine learning task is to discriminate between normal and anomalous data points. In practice, it is not always sufficient to reach high accuracy at this task, one also would like to understand why a given data point has been predicted in a certain way. We present a new principled approach for one-class SVMs that decomposes outlier predictions in terms of input variables. The method first recomposes the one-class model as a neural network with distance functions and min-pooling, and then performs a deep Taylor decomposition (DTD) of the model output. The proposed One-Class DTD is applicable to a number of common distance-based SVM kernels and is able to reliably explain a wide set of data anomalies. Furthermore, it outperforms baselines such as sensitivity analysis, nearest neighbor, or simple edge detection.