Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929295902900224 |
|---|---|
| author | Zhao, Shuai Gan, Leilei Tuan, Luu Anh Fu, Jie Lyu, Lingjuan Jia, Meihuizi Wen, Jinming |
| author_facet | Zhao, Shuai Gan, Leilei Tuan, Luu Anh Fu, Jie Lyu, Lingjuan Jia, Meihuizi Wen, Jinming |
| contents | Recently, various parameter-efficient fine-tuning (PEFT) strategies for application to language models have been proposed and successfully implemented. However, this raises the question of whether PEFT, which only updates a limited set of model parameters, constitutes security vulnerabilities when confronted with weight-poisoning backdoor attacks. In this study, we show that PEFT is more susceptible to weight-poisoning backdoor attacks compared to the full-parameter fine-tuning method, with pre-defined triggers remaining exploitable and pre-defined targets maintaining high confidence, even after fine-tuning. Motivated by this insight, we developed a Poisoned Sample Identification Module (PSIM) leveraging PEFT, which identifies poisoned samples through confidence, providing robust defense against weight-poisoning backdoor attacks. Specifically, we leverage PEFT to train the PSIM with randomly reset sample labels. During the inference process, extreme confidence serves as an indicator for poisoned samples, while others are clean. We conduct experiments on text classification tasks, five fine-tuning strategies, and three weight-poisoning backdoor attack methods. Experiments show near 100% success rates for weight-poisoning backdoor attacks when utilizing PEFT. Furthermore, our defensive approach exhibits overall competitive performance in mitigating weight-poisoning backdoor attacks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_12168 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning Zhao, Shuai Gan, Leilei Tuan, Luu Anh Fu, Jie Lyu, Lingjuan Jia, Meihuizi Wen, Jinming Cryptography and Security Artificial Intelligence Computation and Language Recently, various parameter-efficient fine-tuning (PEFT) strategies for application to language models have been proposed and successfully implemented. However, this raises the question of whether PEFT, which only updates a limited set of model parameters, constitutes security vulnerabilities when confronted with weight-poisoning backdoor attacks. In this study, we show that PEFT is more susceptible to weight-poisoning backdoor attacks compared to the full-parameter fine-tuning method, with pre-defined triggers remaining exploitable and pre-defined targets maintaining high confidence, even after fine-tuning. Motivated by this insight, we developed a Poisoned Sample Identification Module (PSIM) leveraging PEFT, which identifies poisoned samples through confidence, providing robust defense against weight-poisoning backdoor attacks. Specifically, we leverage PEFT to train the PSIM with randomly reset sample labels. During the inference process, extreme confidence serves as an indicator for poisoned samples, while others are clean. We conduct experiments on text classification tasks, five fine-tuning strategies, and three weight-poisoning backdoor attack methods. Experiments show near 100% success rates for weight-poisoning backdoor attacks when utilizing PEFT. Furthermore, our defensive approach exhibits overall competitive performance in mitigating weight-poisoning backdoor attacks. |
| title | Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning |
| topic | Cryptography and Security Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2402.12168 |