Anomaly-Injected Deep Support Vector Data Description for Text Outlier Detection

TL;DR

This paper introduces AI-SVDD, a novel deep learning framework that enhances textual anomaly detection by combining support vector data description with anomaly injection and BERT-based text representations.

Contribution

The work presents a new deep anomaly detection method for text that integrates anomaly injection with support vector data description and leverages BERT for improved representation.

Findings

AI-SVDD outperforms existing methods on multiple datasets

Incorporating known anomalies improves detection accuracy

Combines MLP and BERT for effective text feature extraction

Abstract

Anomaly detection or outlier detection is a common task in various domains, which has attracted significant research efforts in recent years. Existing works mainly focus on structured data such as numerical or categorical data; however, anomaly detection on unstructured textual data is less attended. In this work, we target the textual anomaly detection problem and propose a deep anomaly-injected support vector data description (AI-SVDD) framework. AI-SVDD not only learns a more compact representation of the data hypersphere but also adopts a small number of known anomalies to increase the discriminative power. To tackle text input, we employ a multilayer perceptron (MLP) network in conjunction with BERT to obtain enriched text representations. We conduct experiments on three text anomaly detection applications with multiple datasets. Experimental results show that the proposed AI-SVDD is promising and outperforms existing works.