Flow-based SVDD for anomaly detection

TL;DR

FlowSVDD introduces a flow-based deep learning model for anomaly detection that maintains a well-formed hypersphere, achieving competitive results and outperforming existing deep SVDD methods on benchmarks.

Contribution

The paper presents FlowSVDD, a novel flow-based deep SVDD model that prevents hypersphere collapse and improves anomaly detection performance.

Findings

Achieves comparable results to state-of-the-art methods

Significantly outperforms related deep SVDD methods on benchmarks

Maintains a well-formed hypersphere in deep SVDD

Abstract

We propose FlowSVDD -- a flow-based one-class classifier for anomaly/outliers detection that realizes a well-known SVDD principle using deep learning tools. Contrary to other approaches to deep SVDD, the proposed model is instantiated using flow-based models, which naturally prevents from collapsing of bounding hypersphere into a single point. Experiments show that FlowSVDD achieves comparable results to the current state-of-the-art methods and significantly outperforms related deep SVDD methods on benchmark datasets.