Rethinking Contrastive Learning in Graph Anomaly Detection: A Clean-View Perspective

TL;DR

This paper introduces CVGAD, a novel graph anomaly detection framework that improves contrastive learning by identifying and removing interfering edges, leading to more accurate detection of anomalies in graph data.

Contribution

The paper proposes a multi-scale anomaly awareness and progressive purification modules to enhance contrastive learning in graph anomaly detection, addressing interference issues.

Findings

Outperforms existing methods on five benchmark datasets.

Effectively identifies and removes interfering edges.

Improves anomaly detection accuracy significantly.

Abstract

Graph anomaly detection aims to identify unusual patterns in graph-based data, with wide applications in fields such as web security and financial fraud detection. Existing methods typically rely on contrastive learning, assuming that a lower similarity between a node and its local subgraph indicates abnormality. However, these approaches overlook a crucial limitation: the presence of interfering edges invalidates this assumption, since it introduces disruptive noise that compromises the contrastive learning process. Consequently, this limitation impairs the ability to effectively learn meaningful representations of normal patterns, leading to suboptimal detection performance. To address this issue, we propose a Clean-View Enhanced Graph Anomaly Detection framework (CVGAD), which includes a multi-scale anomaly awareness module to identify key sources of interference in the contrastive learning process. Moreover, to mitigate bias from the one-step edge removal process, we introduce a novel progressive purification module. This module incrementally refines the graph by iteratively identifying and removing interfering edges, thereby enhancing model performance. Extensive experiments on five benchmark datasets validate the effectiveness of our approach.