SEEP: Training Dynamics Grounds Latent Representation Search for Mitigating Backdoor Poisoning Attacks
This work addresses a critical security issue for NLP models trained on public datasets, offering improved protection against advanced backdoor attacks compared to existing methods.
The paper tackles the problem of defending NLP models against advanced backdoor poisoning attacks by proposing a novel defensive mechanism that uses training dynamics to identify poisoned samples with high precision and label propagation to improve recall, resulting in a significant reduction in attack success rates while maintaining high classification accuracy on clean test sets.
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.