Semi-supervised Graph Anomaly Detection via Robust Homophily Learning

TL;DR

This paper introduces RHO, a novel semi-supervised graph anomaly detection method that adaptively learns diverse homophily patterns in normal nodes, overcoming limitations of previous assumptions and improving detection accuracy.

Contribution

RHO proposes adaptive spectral filters and a homophily alignment mechanism to robustly learn normal node patterns with diverse homophily in semi-supervised GAD.

Findings

RHO outperforms state-of-the-art methods on eight real-world datasets.

It effectively captures diverse and under-represented homophily patterns.

The approach demonstrates significant improvements in anomaly detection accuracy.

Abstract

Semi-supervised graph anomaly detection (GAD) utilizes a small set of labeled normal nodes to identify abnormal nodes from a large set of unlabeled nodes in a graph. Current methods in this line posit that 1) normal nodes share a similar level of homophily and 2) the labeled normal nodes can well represent the homophily patterns in the normal class. However, this assumption often does not hold well since normal nodes in a graph can exhibit diverse homophily in real-world GAD datasets. In this paper, we propose RHO, namely Robust Homophily Learning, to adaptively learn such homophily patterns. RHO consists of two novel modules, adaptive frequency response filters (AdaFreq) and graph normality alignment (GNA). AdaFreq learns a set of adaptive spectral filters that capture different frequency components of the labeled normal nodes with varying homophily in the channel-wise and cross-channel views of node attributes. GNA is introduced to enforce consistency between the channel-wise and cross-channel homophily representations to robustify the normality learned by the filters in the two views. Experiments on eight real-world GAD datasets show that RHO can effectively learn varying, often under-represented, homophily in the small normal node set and substantially outperforms state-of-the-art competing methods. Code is available at https://github.com/mala-lab/RHO.