Class Label-aware Graph Anomaly Detection

TL;DR

This paper introduces CLAD, a novel framework that leverages limited class labels to improve unsupervised graph anomaly detection, demonstrating superior performance across multiple datasets.

Contribution

The work proposes a class label-aware approach for graph anomaly detection that effectively utilizes limited labeled data to enhance detection accuracy.

Findings

CLAD outperforms existing unsupervised GAD methods on ten datasets.

Limited class labels significantly improve anomaly detection performance.

The approach is effective even without ground-truth class label information.

Abstract

Unsupervised GAD methods assume the lack of anomaly labels, i.e., whether a node is anomalous or not. One common observation we made from previous unsupervised methods is that they not only assume the absence of such anomaly labels, but also the absence of class labels (the class a node belongs to used in a general node classification task). In this work, we study the utility of class labels for unsupervised GAD; in particular, how they enhance the detection of structural anomalies. To this end, we propose a Class Label-aware Graph Anomaly Detection framework (CLAD) that utilizes a limited amount of labeled nodes to enhance the performance of unsupervised GAD. Extensive experiments on ten datasets demonstrate the superior performance of CLAD in comparison to existing unsupervised GAD methods, even in the absence of ground-truth class label information. The source code for CLAD is available at \url{https://github.com/jhkim611/CLAD}.