Graph Injection Attack

Graph injection attacks (GIAs) target the vulnerability of graph neural networks (GNNs) by introducing malicious nodes into the graph structure, aiming to manipulate GNN predictions. Current research focuses on developing more sophisticated attack strategies, including text-level injections and reinforcement learning-based approaches to optimize attack effectiveness, while simultaneously exploring robust defense mechanisms such as certified robustness frameworks and homophily-based augmentation. Understanding and mitigating GIAs is crucial for ensuring the reliability and security of GNNs across diverse applications, from recommendation systems to social network analysis.

Papers

May 26, 2024

Intruding with Words: Towards Understanding Graph Injection Attacks at the Text Level
Runlin Lei, Yuwei Hu, Yuchen Ren, Zhewei Wei
Adversarial Attack Word List Node Embeddings Graph Injection Attack

March 3, 2024

Collective Certified Robustness against Graph Injection Attacks
Yuni Lai, Bailin Pan, Kaihuang Chen, Yancheng Yuan, Kai Zhou
Native Robustness Linear Programming Graph Injection Attack Sample Wise Certificate Collective Robustness

December 7, 2023

Node-aware Bi-smoothing: Certified Robustness against Graph Injection Attacks
Yuni Lai, Yulin Zhu, Bailin Pan, Kai Zhou
Native Robustness Node Aware Bi Smoothing Graph Injection Attack

May 4, 2023

Single Node Injection Label Specificity Attack on Graph Neural Networks via Reinforcement Learning
Dayuan Chen, Jian Zhang, Yuqian Lv, Jinhuan Wang, Hongjie Ni, Shanqing Yu, Zhen Wang, Qi Xuan
Reinforcement Learning Graph Neural Network Gradient Based Adversarial Label Attack Node Injection Graph Injection Attack

November 15, 2022

Resisting Graph Adversarial Attack via Cooperative Homophilous Augmentation
Zhihao Zhu, Chenwang Wu, Min Zhou, Hao Liao, Defu Lian, Enhong Chen
Adversarial Attack Random Node State of the Art Defense Defense Framework Graph Adversarial Attack Graph Injection Attack Collaborative Augmentation

October 25, 2022

Motif-Backdoor: Rethinking the Backdoor Attack on Graph Neural Networks via Motifs
Haibin Zheng, Haiyang Xiong, Jinyin Chen, Haonan Ma, Guohan Huang
Graph Neural Network Backdoor Attack Jamdani Motif Graph Injection Attack Motif Based Graph

May 19, 2022

A Simple Yet Effective SVD-GCN for Directed Graphs
Chunya Zou, Andi Han, Lequan Lin, Junbin Gao
Singular Value Decomposition Directed Graph Graph Signal Framelet Transform Graph Injection Attack Graph Framelets

February 16, 2022

Understanding and Improving Graph Injection Attack by Promoting Unnoticeability
Yongqiang Chen, Han Yang, Yonggang Zhang, Kaili Ma, Tongliang Liu, Bo Han, James Cheng
Graph Neural Network Human Understanding Adversarial Objective Information Theoretic Detectability Graph Injection Attack

January 27, 2022

FairEGM: Fair Link Prediction and Recommendation via Emulated Graph Modification
Sean Current, Yuntian He, Saket Gurukar, Srinivasan Parthasarathy
Graph Neural Network Link Prediction Effective Recommendation Inherent Bias Recommendation Fairness Graph Injection Attack Fairness Optimization

January 17, 2022

GCS: Graph-based Coordination Strategy for Multi-Agent Reinforcement Learning
Jingqing Ruan, Yali Du, Xuantang Xiong, Dengpeng Xing, Xiyun Li, Linghui Meng, Haifeng Zhang, Jun Wang, Bo Xu
Multi Agent Reinforcement Learning Graph Based Diverse Policy Decentralized Control Dynamic Decision Coordination Graph Graph Injection Attack