Topology Attack

Topology attacks exploit the vulnerabilities of graph neural networks (GNNs) by manipulating the graph structure, aiming to misclassify nodes or otherwise compromise the GNN's performance. Current research focuses on developing both more effective attack strategies, such as those employing adaptive budget allocation and dynamic optimization algorithms, and robust detection methods that leverage the inherent message-passing mechanisms of GNNs to identify manipulated nodes. Understanding and mitigating these attacks is crucial for ensuring the reliability and security of GNNs in various applications, ranging from social network analysis to fraud detection.

Papers

March 5, 2024

Minimum Topology Attacks for Graph Neural Networks
Mengmei Zhang, Xiao Wang, Chuan Shi, Lingjuan Lyu, Tianchi Yang, Junping Du
Graph Neural Network Topology Attack

February 21, 2024

Verifying message-passing neural networks via topology-based bounds tightening
Christopher Hojny, Shiqiang Zhang, Juan S. Campos, Ruth Misener
Verification Task Message Passing Neural Network Upper Bound Topology Attack

November 28, 2022

ARISE: Graph Anomaly Detection on Attributed Networks via Substructure Awareness
Jingcan Duan, Bin Xiao, Siwei Wang, Haifang Zhou, Xinwang Liu
Graph Anomaly Detection Attributed Network Network Anomaly Detection Anomaly Node ApeRIodic SEmi Parametric Topology Attack

April 21, 2022

Detecting Topology Attacks against Graph Neural Networks
Senrong Xu, Yuan Yao, Liangyue Li, Wei Yang, Feng Xu, Hanghang Tong
Graph Neural Network Node Classification Topology Attack Victim Node

Topology Attack

Papers

Minimum Topology Attacks for Graph Neural Networks

Verifying message-passing neural networks via topology-based bounds tightening

ARISE: Graph Anomaly Detection on Attributed Networks via Substructure Awareness

Detecting Topology Attacks against Graph Neural Networks