Graph Contrastive Backdoor Attacks

Hangfan Zhang, Jinghui Chen, Lu Lin, Jinyuan Jia, Dinghao Wu
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:40888-40910, 2023.

Abstract

Graph Contrastive Learning (GCL) has attracted considerable interest due to its impressive node representation learning capability. Despite the wide application of GCL techniques, little attention has been paid to the security of GCL. In this paper, we systematically study the vulnerability of GCL in the presence of malicious backdoor adversaries. In particular, we propose GCBA, the first backdoor attack for graph contrastive learning. GCBA incorporates three attacks: poisoning, crafting, and natural backdoor, each targeting one stage of the GCL pipeline. We formulate our attacks as optimization problems and solve them with a novel discrete optimization technique to overcome the discrete nature of graph-structured data. By extensively evaluating GCBA on multiple datasets and GCL methods, we show that our attack can achieve high attack success rates while preserving stealthiness. We further consider potential countermeasures to our attack and conclude that existing defenses are insufficient to mitigate GCBA. We show that as a complex paradigm involving data and model republishing, GCL is vulnerable to backdoor attacks, and specifically designed defenses are needed to mitigate the backdoor attacks on GCL.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-zhang23e, title = {Graph Contrastive Backdoor Attacks}, author = {Zhang, Hangfan and Chen, Jinghui and Lin, Lu and Jia, Jinyuan and Wu, Dinghao}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {40888--40910}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/zhang23e/zhang23e.pdf}, url = {https://proceedings.mlr.press/v202/zhang23e.html}, abstract = {Graph Contrastive Learning (GCL) has attracted considerable interest due to its impressive node representation learning capability. Despite the wide application of GCL techniques, little attention has been paid to the security of GCL. In this paper, we systematically study the vulnerability of GCL in the presence of malicious backdoor adversaries. In particular, we propose GCBA, the first backdoor attack for graph contrastive learning. GCBA incorporates three attacks: poisoning, crafting, and natural backdoor, each targeting one stage of the GCL pipeline. We formulate our attacks as optimization problems and solve them with a novel discrete optimization technique to overcome the discrete nature of graph-structured data. By extensively evaluating GCBA on multiple datasets and GCL methods, we show that our attack can achieve high attack success rates while preserving stealthiness. We further consider potential countermeasures to our attack and conclude that existing defenses are insufficient to mitigate GCBA. We show that as a complex paradigm involving data and model republishing, GCL is vulnerable to backdoor attacks, and specifically designed defenses are needed to mitigate the backdoor attacks on GCL.} }
Endnote
%0 Conference Paper %T Graph Contrastive Backdoor Attacks %A Hangfan Zhang %A Jinghui Chen %A Lu Lin %A Jinyuan Jia %A Dinghao Wu %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-zhang23e %I PMLR %P 40888--40910 %U https://proceedings.mlr.press/v202/zhang23e.html %V 202 %X Graph Contrastive Learning (GCL) has attracted considerable interest due to its impressive node representation learning capability. Despite the wide application of GCL techniques, little attention has been paid to the security of GCL. In this paper, we systematically study the vulnerability of GCL in the presence of malicious backdoor adversaries. In particular, we propose GCBA, the first backdoor attack for graph contrastive learning. GCBA incorporates three attacks: poisoning, crafting, and natural backdoor, each targeting one stage of the GCL pipeline. We formulate our attacks as optimization problems and solve them with a novel discrete optimization technique to overcome the discrete nature of graph-structured data. By extensively evaluating GCBA on multiple datasets and GCL methods, we show that our attack can achieve high attack success rates while preserving stealthiness. We further consider potential countermeasures to our attack and conclude that existing defenses are insufficient to mitigate GCBA. We show that as a complex paradigm involving data and model republishing, GCL is vulnerable to backdoor attacks, and specifically designed defenses are needed to mitigate the backdoor attacks on GCL.
APA
Zhang, H., Chen, J., Lin, L., Jia, J. & Wu, D.. (2023). Graph Contrastive Backdoor Attacks. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:40888-40910 Available from https://proceedings.mlr.press/v202/zhang23e.html.

Related Material