Double-Adversarial Activation Anomaly Detection: Adversarial Autoencoders are Anomaly Generators

TL;DR

DA3D introduces a novel unsupervised anomaly detection method that uses adversarial autoencoders to generate artificial anomalies, transforming the task into a supervised learning problem and outperforming existing methods without domain knowledge.

Contribution

The paper presents a new generative approach using adversarial autoencoders to create artificial anomalies for improved unsupervised anomaly detection.

Findings

DA3D outperforms state-of-the-art methods in anomaly detection.

Artificial anomalies improve detection of unseen real anomalies.

No domain knowledge needed for effective detection.

Abstract

Anomaly detection is a challenging task for machine learning algorithms due to the inherent class imbalance. It is costly and time-demanding to manually analyse the observed data, thus usually only few known anomalies if any are available. Inspired by generative models and the analysis of the hidden activations of neural networks, we introduce a novel unsupervised anomaly detection method called DA3D. Here, we use adversarial autoencoders to generate anomalous counterexamples based on the normal data only. These artificial anomalies used during training allow the detection of real, yet unseen anomalies. With our novel generative approach, we transform the unsupervised task of anomaly detection to a supervised one, which is more tractable by machine learning and especially deep learning methods. DA3D surpasses the performance of state-of-the-art anomaly detection methods in a purely data-driven way, where no domain knowledge is required.