Understanding the limitations of self-supervised learning for tabular anomaly detection
This work addresses a gap in anomaly detection for tabular data, showing incremental insights into the limitations of self-supervised methods.
The paper tackled the problem of applying self-supervised learning to tabular anomaly detection, finding that it does not improve performance compared to raw data due to neural networks introducing irrelevant features, but performance can be recovered by using a subspace of the representation.
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.