# Autoencoders for Anomaly Detection are Unreliable

arXiv: https://arxiv.org/abs/2501.13864 · HTML: https://mlsanity.com/paper/2501.13864/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 37 references that are in the map.
- First posted on arXiv 23 Jan 2025.

## Builds on (most cited first, 16 of 16)

- [Towards Deep Learning Models Resistant to Adversarial Attacks](https://mlsanity.com/paper/1706.06083/) (2017)
- [Real-world Anomaly Detection in Surveillance Videos](https://mlsanity.com/paper/1801.04264/) (2018)
- [Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detection](https://mlsanity.com/paper/1904.02639/) (2019)
- [Do Deep Generative Models Know What They Don't Know?](https://mlsanity.com/paper/1810.09136/) (2018)
- [A Unified Model for Multi-class Anomaly Detection](https://mlsanity.com/paper/2206.03687/) (2022)
- [Improving Reconstruction Autoencoder Out-of-distribution Detection with Mahalanobis Distance](https://mlsanity.com/paper/1812.02765/) (2018)
- [Hierarchical Vector Quantized Transformer for Multi-class Unsupervised Anomaly Detection](https://mlsanity.com/paper/2310.14228/) (2023)
- [Rethinking Reconstruction Autoencoder-Based Out-of-Distribution Detection](https://mlsanity.com/paper/2203.02194/) (2022)
- [Hierarchical VAEs Know What They Don't Know](https://mlsanity.com/paper/2102.08248/) (2021)
- [Robust Anomaly Detection in Images using Adversarial Autoencoders](https://mlsanity.com/paper/1901.06355/) (2019)
- [Learning Not to Reconstruct Anomalies](https://mlsanity.com/paper/2110.09742/) (2021)
- [ARAE: Adversarially Robust Training of Autoencoders Improves Novelty Detection](https://mlsanity.com/paper/2003.05669/) (2020)
- [Autoencoding Under Normalization Constraints](https://mlsanity.com/paper/2105.05735/) (2021)
- [Rethinking Autoencoders for Medical Anomaly Detection from A Theoretical Perspective](https://mlsanity.com/paper/2403.09303/) (2024)
- [Unsupervised anomaly detection algorithms on real-world data: how many do we need?](https://mlsanity.com/paper/2305.00735/) (2023)
- [Exploiting Autoencoder's Weakness to Generate Pseudo Anomalies](https://mlsanity.com/paper/2405.05886/) (2024)

## Built on by (most cited first, 9 of 9)

- [Failure Identification in Imitation Learning Via Statistical and Semantic Filtering](https://mlsanity.com/paper/2604.13788/) (2026)
- [Toward Faithful Explanations in Acoustic Anomaly Detection](https://mlsanity.com/paper/2601.12660/) (2026)
- [Risk-Based Thresholding for Reliable Anomaly Detection in Concentrated Solar Power Plants](https://mlsanity.com/paper/2503.19146/) (2025)
- [Applying Deep Learning to Anomaly Detection of Russian Satellite Activity for Indications Prior to Military Activity](https://mlsanity.com/paper/2509.00050/) (2025)
- [Enhancing ECG Classification Robustness with Lightweight Unsupervised Anomaly Detection Filters](https://mlsanity.com/paper/2510.26501/) (2025)
- [Hybrid Quantum-Classical Autoencoders for Unsupervised Network Intrusion Detection](https://mlsanity.com/paper/2512.05069/) (2025)
- [Modeling Normal Is All You Need: Joint Latent Clustering for Anomaly Detection in Multimodal Cyber-Physical Systems](https://mlsanity.com/paper/2607.06094/) (2026)
- [Cross-Domain Industrial Fault Detection by Causal Mechanism Monitoring](https://mlsanity.com/paper/2608.14666/) (2026)
- [When Does Unsupervised Learning Succeed or Fail? A PoS Perspective on Reconstruction-Based Anomaly Detection](https://mlsanity.com/paper/2609.28832/) (2026)

## Abstract (arXiv)

Autoencoders are frequently used for anomaly detection, both in the unsupervised and semi-supervised settings. They rely on the assumption that when trained using the reconstruction loss, they will be able to reconstruct normal data more accurately than anomalous data. Some recent works have posited that this assumption may not always hold, but little has been done to study the validity of the assumption in theory. In this work we show that this assumption indeed does not hold, and illustrate that anomalies, lying far away from normal data, can be perfectly reconstructed in practice. We revisit the theory of failure of linear autoencoders for anomaly detection by showing how they can perfectly reconstruct out of bounds, or extrapolate undesirably, and note how this can be dangerous in safety critical applications. We connect this to non-linear autoencoders through experiments on both tabular data and real-world image data, the two primary application areas of autoencoders for anomaly detection.