Mitigating Homophily Disparity in Graph Anomaly Detection: A Scalable and Adaptive Approach

2026-03-09

AI Summary

Graph anomaly detection faces challenges arising from both node-level and class-level homophily discrepancies, as well as limited scalability. This work proposes SAGAD, a novel framework that simultaneously mitigates these two types of homophily issues for the first time in graph anomaly detection. SAGAD efficiently captures both high- and low-frequency signals through precomputed multi-hop embeddings and a reparameterized Chebyshev filter. It further introduces an anomaly-context adaptive fusion mechanism based on the Rayleigh quotient and a frequency-preference-guided loss to ensure asymptotic linear separability between normal and anomalous nodes. Combined with mini-batch training, SAGAD achieves state-of-the-art performance across ten benchmark datasets, offering high accuracy alongside linear time and space complexity, thereby substantially reducing memory consumption on large-scale graphs.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Multiagent Systems: Adversarial Agents

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Graph anomaly detection (GAD) aims to identify nodes that deviate from normal patterns in structure or features. While recent GNN-based approaches have advanced this task, they struggle with two major challenges: 1) homophily disparity, where nodes exhibit varying homophily at both class and node levels; and 2) limited scalability, as many methods rely on costly whole-graph operations. To address them, we propose SAGAD, a Scalable and Adaptive framework for GAD. SAGAD precomputes multi-hop embeddings and applies reparameterized Chebyshev filters to extract low- and high-frequency information, enabling efficient training and capturing both homophilic and heterophilic patterns. To mitigate node-level homophily disparity, we introduce an Anomaly Context-Aware Adaptive Fusion, which adaptively fuses low- and high-pass embeddings using fusion coefficients conditioned on Rayleigh Quotient-guided anomalous subgraph structures for each node. To alleviate class-level disparity, we design a Frequency Preference Guidance Loss, which encourages anomalies to preserve more high-frequency information than normal nodes. SAGAD supports mini-batch training, achieves linear time and space complexity, and drastically reduces memory usage on large-scale graphs. Theoretically, SAGAD ensures asymptotic linear separability between normal and abnormal nodes under mild conditions. Extensive experiments on 10 benchmarks confirm SAGAD's superior accuracy and scalability over state-of-the-art methods.

Problem

• graph anomaly detection
• homophily disparity
• scalability
• node-level homophily
• class-level homophily

Innovation

• Graph Anomaly Detection
• Homophily Disparity
• Adaptive Fusion
• Scalable GNN
• Frequency-aware Learning