Disentangling Tabular Data Towards Better One-Class Anomaly Detection

TL;DR

This paper introduces a novel disentanglement approach for one-class anomaly detection in tabular data, effectively capturing attribute correlations to improve detection accuracy over existing methods.

Contribution

It proposes a new method that disentangles attribute subsets (CorrSets) to better learn normal data correlations, pioneering the use of disentanglement in tabular one-class anomaly detection.

Findings

Outperforms state-of-the-art methods on 20 datasets

Achieves 6.1% higher AUC-PR on average

Achieves 2.1% higher AUC-ROC on average

Abstract

Tabular anomaly detection under the one-class classification setting poses a significant challenge, as it involves accurately conceptualizing "normal" derived exclusively from a single category to discern anomalies from normal data variations. Capturing the intrinsic correlation among attributes within normal samples presents one promising method for learning the concept. To do so, the most recent effort relies on a learnable mask strategy with a reconstruction task. However, this wisdom may suffer from the risk of producing uniform masks, i.e., essentially nothing is masked, leading to less effective correlation learning. To address this issue, we presume that attributes related to others in normal samples can be divided into two non-overlapping and correlated subsets, defined as CorrSets, to capture the intrinsic correlation effectively. Accordingly, we introduce an innovative method that disentangles CorrSets from normal tabular data. To our knowledge, this is a pioneering effort to apply the concept of disentanglement for one-class anomaly detection on tabular data. Extensive experiments on 20 tabular datasets show that our method substantially outperforms the state-of-the-art methods and leads to an average performance improvement of 6.1% on AUC-PR and 2.1% on AUC-ROC. Codes are available at https://github.com/yjnanan/Disent-AD.