Backdoor Attacks to Graph Neural Networks
This addresses security vulnerabilities in GNNs, which are critical for applications like social network analysis and bioinformatics, but the work is incremental as it adapts existing backdoor attack concepts to a new domain.
The authors tackled the problem of backdoor attacks on graph neural networks (GNNs) for graph classification by proposing a subgraph-based method, achieving effective attacks with minimal impact on clean graph accuracy across three real-world datasets.
In this work, we propose the first backdoor attack to graph neural networks (GNN). Specifically, we propose a \emph{subgraph based backdoor attack} to GNN for graph classification. In our backdoor attack, a GNN classifier predicts an attacker-chosen target label for a testing graph once a predefined subgraph is injected to the testing graph. Our empirical results on three real-world graph datasets show that our backdoor attacks are effective with a small impact on a GNN's prediction accuracy for clean testing graphs. Moreover, we generalize a randomized smoothing based certified defense to defend against our backdoor attacks. Our empirical results show that the defense is effective in some cases but ineffective in other cases, highlighting the needs of new defenses for our backdoor attacks.