SEEP: Training Dynamics Grounds Latent Representation Search for Mitigating Backdoor Poisoning Attacks

Xuanli He, Qiongkai Xu, Jun Wang, Benjamin I. P. Rubinstein, Trevor Cohn


Abstract
Modern NLP models are often trained on public datasets drawn from diverse sources, rendering them vulnerable to data poisoning attacks. These attacks can manipulate the model’s behavior in ways engineered by the attacker. One such tactic involves the implantation of backdoors, achieved by poisoning specific training instances with a textual trigger and a target class label. Several strategies have been proposed to mitigate the risks associated with backdoor attacks by identifying and removing suspected poisoned examples. However, we observe that these strategies fail to offer effective protection against several advanced backdoor attacks. To remedy this deficiency, we propose a novel defensive mechanism that first exploits training dynamics to identify poisoned samples with high precision, followed by a label propagation step to improve recall and thus remove the majority of poisoned instances. Compared with recent advanced defense methods, our method considerably reduces the success rates of several backdoor attacks while maintaining high classification accuracy on clean test sets.
Anthology ID:
2024.tacl-1.55
Volume:
Transactions of the Association for Computational Linguistics, Volume 12
Month:
Year:
2024
Address:
Cambridge, MA
Venue:
TACL
SIG:
Publisher:
MIT Press
Note:
Pages:
996–1010
Language:
URL:
https://aclanthology.org/2024.tacl-1.55
DOI:
10.1162/tacl_a_00684
Bibkey:
Cite (ACL):
Xuanli He, Qiongkai Xu, Jun Wang, Benjamin I. P. Rubinstein, and Trevor Cohn. 2024. SEEP: Training Dynamics Grounds Latent Representation Search for Mitigating Backdoor Poisoning Attacks. Transactions of the Association for Computational Linguistics, 12:996–1010.
Cite (Informal):
SEEP: Training Dynamics Grounds Latent Representation Search for Mitigating Backdoor Poisoning Attacks (He et al., TACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.tacl-1.55.pdf