NeurIPS 2022

Timezone: »

Poster

Dual-discriminative Graph Neural Network for Imbalanced Graph-level Anomaly Detection

GE ZHANG · Zhenyu Yang · Jia Wu · Jian Yang · Shan Xue · Hao Peng · Jianlin Su · Chuan Zhou · Quan Z. Sheng · Leman Akoglu · Charu Aggarwal

Tue Nov 29 02:00 PM -- 04:00 PM (PST) @ Hall J #629

in Poster Session 2 »

Graph-level anomaly detection aims to distinguish anomalous graphs in a graph dataset from normal graphs. Anomalous graphs represent a very few but essential patterns in the real world. The anomalous property of a graph may be referable to its anomalous attributes of particular nodes and anomalous substructures that refer to a subset of nodes and edges in the graph. In addition, due to the imbalance nature of anomaly problem, anomalous information will be diluted by normal graphs with overwhelming quantities. Various anomaly notions in the attributes and/or substructures and the imbalance nature together make detecting anomalous graphs a non-trivial task. In this paper, we propose a graph neural network for graph-level anomaly detection, namely iGAD. Specifically, an anomalous graph attribute-aware graph convolution and an anomalous graph substructure-aware deep Random Walk Kernel (deep RWK) are welded into a graph neural network to achieve the dual-discriminative ability on anomalous attributes and substructures. Deep RWK in iGAD makes up for the deficiency of graph convolution in distinguishing structural information caused by the simple neighborhood aggregation mechanism. Further, we propose a Point Mutual Information (PMI)-based loss function to target the problems caused by imbalance distributions. PMI-based loss function enables iGAD to capture essential correlation between input graphs and their anomalous/normal properties. We evaluate iGAD on four real-world graph datasets. Extensive experiments demonstrate the superiority of iGAD on the graph-level anomaly detection task.

Keywords: graph neural networks · Imbalanced data · Graph-level Anomaly Detection

[ Paper] [ Poster] [ OpenReview]

Author Information

GE ZHANG (Macquarie University)

Zhenyu Yang (Macquarie University)

Jia Wu (Macquarie University)

Jian Yang (Macquarie University)

Shan Xue (Macquarie University)

Hao Peng (Beihang University)

Jianlin Su (Shenzhen Zhuiyi Technology Co., Ltd.)

Chuan Zhou (Chinese Academy of Sciences)

Quan Z. Sheng (Macquarie University)

Dr. Michael Sheng is a full Professor and Head of School of Computing at Macquarie University. Before moving to Macquarie, Michael spent 10 years at School of Computer Science, the University of Adelaide (UoA), serving in a number of senior leadership roles including acting Head and Deputy Head of School of Computer Science. Michael holds a PhD degree in computer science from the University of New South Wales (UNSW) and did his post-doc as a research scientist at CSIRO ICT Centre. Dr. Michael Sheng is ranked by Microsoft Academic as one of the Most Impactful Authors in Services Computing (ranked Top 5 All Time) and in Web of Things (ranked Top 20 All Time). He is the recipient of the AMiner Most Influential Scholar Award on IoT (2007-2017), ARC Future Fellowship (2014), Chris Wallace Award for Outstanding Research Contribution (2012), and Microsoft Research Fellowship (2003). Prof Michael Sheng is Vice Chair of the Executive Committee of the IEEE Technical Community on Services Computing (IEEE TCSVC), the Associate Director (Smart Technologies) of Macquarie's Smart Green Cities Research Centre, and a member of the ACS Technical Advisory Board on IoT.