MADCluster: Model-agnostic Anomaly Detection with Self-supervised Clustering Network

TL;DR

MADCluster is a versatile, self-supervised clustering framework for anomaly detection that improves performance across different deep learning architectures by addressing the hypersphere collapse problem.

Contribution

It introduces a novel model-agnostic anomaly detection method with a new adaptive loss and continuous clustering, applicable to various architectures.

Findings

Improves anomaly detection performance on four benchmark datasets.

Addresses hypersphere collapse issue in deep anomaly detection.

Demonstrates compatibility and effectiveness across multiple architectures.

Abstract

In this paper, we propose MADCluster, a novel model-agnostic anomaly detection framework utilizing self-supervised clustering. MADCluster is applicable to various deep learning architectures and addresses the 'hypersphere collapse' problem inherent in existing deep learning-based anomaly detection methods. The core idea is to cluster normal pattern data into a 'single cluster' while simultaneously learning the cluster center and mapping data close to this center. Also, to improve expressiveness and enable effective single clustering, we propose a new 'One-directed Adaptive loss'. The optimization of this loss is mathematically proven. MADCluster consists of three main components: Base Embedder capturing high-dimensional temporal dynamics, Cluster Distance Mapping, and Sequence-wise Clustering for continuous center updates. Its model-agnostic characteristics are achieved by applying various architectures to the Base Embedder. Experiments on four time series benchmark datasets demonstrate that applying MADCluster improves the overall performance of comparative models. In conclusion, the compatibility of MADCluster shows potential for enhancing model performance across various architectures.