Reconciling Attribute and Structural Anomalies for Improved Graph Anomaly Detection

TL;DR

TripleAD is a novel graph anomaly detection framework that effectively identifies attribute, structural, and mixed anomalies by using three specialized modules and mutual distillation, outperforming existing methods.

Contribution

The paper introduces TripleAD, a triple-channel framework with mutual distillation to better detect various anomaly types in graphs, addressing the limitations of previous single-model approaches.

Findings

TripleAD outperforms strong baselines in detecting graph anomalies.

The multiscale attribute estimation improves node interaction capture.

The link-enhanced structure module aids in identifying isolated nodes.

Abstract

Graph anomaly detection is critical in domains such as healthcare and economics, where identifying deviations can prevent substantial losses. Existing unsupervised approaches strive to learn a single model capable of detecting both attribute and structural anomalies. However, they confront the tug-of-war problem between two distinct types of anomalies, resulting in suboptimal performance. This work presents TripleAD, a mutual distillation-based triple-channel graph anomaly detection framework. It includes three estimation modules to identify the attribute, structural, and mixed anomalies while mitigating the interference between different types of anomalies. In the first channel, we design a multiscale attribute estimation module to capture extensive node interactions and ameliorate the over-smoothing issue. To better identify structural anomalies, we introduce a link-enhanced structure estimation module in the second channel that facilitates information flow to topologically isolated nodes. The third channel is powered by an attribute-mixed curvature, a new indicator that encapsulates both attribute and structural information for discriminating mixed anomalies. Moreover, a mutual distillation strategy is introduced to encourage communication and collaboration between the three channels. Extensive experiments demonstrate the effectiveness of the proposed TripleAD model against strong baselines.