Understanding the limitations of self-supervised learning for tabular anomaly detection

TL;DR

This paper investigates the effectiveness of self-supervised learning for tabular anomaly detection, revealing its limitations and proposing a subspace approach to improve performance.

Contribution

It systematically analyzes why self-supervision fails in tabular anomaly detection and introduces a subspace method to recover detection performance.

Findings

Self-supervised representations do not outperform raw data in tabular anomaly detection.

Neural networks introduce irrelevant features that hinder detection.

Using a subspace of neural representations can recover detection performance.

Abstract

While self-supervised learning has improved anomaly detection in computer vision and natural language processing, it is unclear whether tabular data can benefit from it. This paper explores the limitations of self-supervision for tabular anomaly detection. We conduct several experiments spanning various pretext tasks on 26 benchmark datasets to understand why this is the case. Our results confirm representations derived from self-supervision do not improve tabular anomaly detection performance compared to using the raw representations of the data. We show this is due to neural networks introducing irrelevant features, which reduces the effectiveness of anomaly detectors. However, we demonstrate that using a subspace of the neural network's representation can recover performance.