What is Wrong with One-Class Anomaly Detection?

TL;DR

This paper introduces a new latent class-conditioned anomaly detection framework that leverages hidden class information to improve detection accuracy in diverse normal sample scenarios, outperforming existing methods.

Contribution

It proposes a confidence-based self-labeling framework tailored for latent multi-class anomaly detection, addressing limitations of traditional one-class approaches.

Findings

Outperforms recent one-class AD methods in multi-class scenarios

Effectively leverages hidden class information to improve detection accuracy

Avoids loose decision regions common in one-class methods

Abstract

From a safety perspective, a machine learning method embedded in real-world applications is required to distinguish irregular situations. For this reason, there has been a growing interest in the anomaly detection (AD) task. Since we cannot observe abnormal samples for most of the cases, recent AD methods attempt to formulate it as a task of classifying whether the sample is normal or not. However, they potentially fail when the given normal samples are inherited from diverse semantic labels. To tackle this problem, we introduce a latent class-condition-based AD scenario. In addition, we propose a confidence-based self-labeling AD framework tailored to our proposed scenario. Since our method leverages the hidden class information, it successfully avoids generating the undesirable loose decision region that one-class methods suffer. Our proposed framework outperforms the recent one-class AD methods in the latent multi-class scenarios.