Graph Backdoor Attack
Graph backdoor attacks exploit vulnerabilities in graph neural networks (GNNs), such as those used in recommendation systems and social network analysis, by injecting malicious triggers into training data to manipulate model predictions. Current research focuses on developing more effective and stealthy attacks, including those that modify only labels or leverage graph condensation techniques, as well as designing robust defenses that can identify and mitigate these attacks without compromising model accuracy. Understanding and addressing these attacks is crucial for ensuring the security and reliability of GNNs in various real-world applications.
Papers
October 18, 2024
October 17, 2024
October 7, 2024
July 3, 2024
July 2, 2024
June 15, 2024
June 14, 2024
May 17, 2024
April 19, 2024
December 30, 2023
August 8, 2023
June 17, 2023
April 5, 2023
February 11, 2023
June 21, 2022