Papers by Yanhao Jia
P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Defending Large Language Models (LLMs) against backdoors has long been trapped in a "cat-and-mouse" dilemma where defenders passively react to ever-shifting attack strategies. |
| Approach: | They propose a general and effective defense algorithm that implants benign triggers to reshape the model’s decision boundary. |
| Outcome: | The proposed defense algorithm can neutralize malicious backdoors while preserving task performance. |
Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent models such as OpenAI o1 and DeepSeek-R1 produce explicit reasoning traces, often via Chain-of-Thought prompting. |
| Approach: | They propose a taxonomy that offers a unified perspective for summarizing existing approaches and categorizing reasoning-based backdoor attacks into associative, passive, and active. |
| Outcome: | The proposed taxonomy categorizes reasoning-based backdoor attacks into associative, passive, and active. |
Uni-Retrieval: A Multi-Style Retrieval Framework for STEM’s Education (2025.acl-long)
Copied to clipboard
| Challenge: | Current retrieval models focus on natural text-image retrieval, which is insufficient for STEM education contexts due to ambiguities in the retrieval process. |
| Approach: | They propose a diverse expression retrieval task tailored to educational scenarios . they extract query style features as prototypes and build a continuously updated Prompt Bank . |
| Outcome: | The proposed model outperforms existing retrieval models in most retrieval tasks. |
Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation (2025.findings-acl)
Copied to clipboard
| Challenge: | Parameter-efficient fine-tuning (PEFT) can bridge the gap between large language models and downstream tasks, but is vulnerable to malicious attacks. |
| Approach: | They propose a weak-to-strong unlearning algorithm based on feature alignment knowledge distillation to defend against backdoor attacks . they first train a small-scale language model through full-parameter fine-tuning to serve as the clean teacher model and then guide the large-scale poisoned student model in unlearning the backdoor. |
| Outcome: | The proposed method can unlearn backdoor features without compromising model performance. |