Reconciling Attribute and Structural Anomalies for Improved Graph Anomaly Detection

2025-06-29

AI Summary

Existing unsupervised graph anomaly detection methods struggle to jointly model attribute and structural anomalies, as mutual interference between these two anomaly types degrades detection performance. To address this, we propose TripleAD—a unified framework featuring three parallel, cooperative detection channels for attribute, structural, and hybrid anomalies. TripleAD introduces a multi-scale attribute estimation module and a link-enhanced structural module, along with a novel attribute-mixed curvature metric to quantify local geometric inconsistency. Furthermore, cross-channel knowledge sharing is achieved via mutual distillation. This design effectively mitigates the over-smoothing problem inherent in graph neural networks (GNNs). Extensive experiments on multiple benchmark datasets demonstrate significant improvements in detection accuracy across all three anomaly categories. Notably, TripleAD is the first approach to achieve decoupled modeling and joint optimization of attribute and structural anomalies within a single, coherent framework.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Bridging structured and unstructured data

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.

Problem

• Detect attribute and structural anomalies in graphs
• Mitigate interference between different anomaly types
• Improve performance via mutual distillation-based framework

Innovation

• Triple-channel framework for anomaly detection
• Mutual distillation strategy for collaboration
• Attribute-mixed curvature for mixed anomalies