@misc{indiciae54100680606f,
  title = {Backdoor Samples Detection Based on Perturbation Discrepancy Consistency in Pre-trained Language Models},
  author = {Zuquan Peng and Jianming Fu and Lixin Zou and Li Zheng and Yanzhen Ren and Guojun Peng},
  year = {2025},
  doi = {10.1016/j.neunet.2025.108025},
  url = {https://arxiv.org/abs/2509.05318},
  note = {Source identifier: 2509.05318}
}