BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914399437979648 |
|---|---|
| author | Zhang, Ruyi Gao, Heng Jian, Songlei Tan, Yusong Zhou, Haifang |
| author_facet | Zhang, Ruyi Gao, Heng Jian, Songlei Tan, Yusong Zhou, Haifang |
| contents | Backdoor attacks compromise model reliability by using triggers to manipulate outputs. Trigger inversion can accurately locate these triggers via a generator and is therefore critical for backdoor defense. However, the discrete nature of text prevents existing noise-based trigger generator from being applied to nature language processing (NLP). To overcome the limitations, we employ the rich knowledge embedded in large language models (LLMs) and propose a Backdoor defender powered by LLM Trigger Generator, termed BadLLM-TG. It is optimized through prompt-driven reinforcement learning, using the victim model's feedback loss as the reward signal. The generated triggers are then employed to mitigate the backdoor via adversarial training. Experiments show that our method reduces the attack success rate by 76.2\% on average, outperforming the second-best defender by 13.7. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_15692 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator Zhang, Ruyi Gao, Heng Jian, Songlei Tan, Yusong Zhou, Haifang Cryptography and Security Artificial Intelligence Backdoor attacks compromise model reliability by using triggers to manipulate outputs. Trigger inversion can accurately locate these triggers via a generator and is therefore critical for backdoor defense. However, the discrete nature of text prevents existing noise-based trigger generator from being applied to nature language processing (NLP). To overcome the limitations, we employ the rich knowledge embedded in large language models (LLMs) and propose a Backdoor defender powered by LLM Trigger Generator, termed BadLLM-TG. It is optimized through prompt-driven reinforcement learning, using the victim model's feedback loss as the reward signal. The generated triggers are then employed to mitigate the backdoor via adversarial training. Experiments show that our method reduces the attack success rate by 76.2\% on average, outperforming the second-best defender by 13.7. |
| title | BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2603.15692 |