Constrained Optimization with Dynamic Bound-scaling for Effective NLP Backdoor Defense

Guangyu Shen, Yingqi Liu, Guanhong Tao, Qiuling Xu, Zhuo Zhang, Shengwei An, Shiqing Ma, Xiangyu Zhang

Abstract: Modern language models are vulnerable to backdoor attacks. An injected malicious token sequence (i.e., a trigger) can cause the compromised model to misbehave, raising security concerns. Trigger in...