Ranjie Duan

Pruning as a Cooperative Game: Surrogate-Assisted Layer Contribution Estimation for Large Language Models

Vulcan: Crafting Compact Class-Specific Vision Transformers For Edge Intelligence

Towards Safe Reasoning in Large Reasoning Models via Corrective Intervention

Mitigating Error Propagation in Low-Rank Approximation of Large Models via Distribution-Aware Whitening

Obscure but Effective: Classical Chinese Jailbreak Prompt Optimization via Bio-Inspired Search

Inverse Reinforcement Learning with Dynamic Reward Scaling for LLM Alignment

AgenticPA: Toward Automated and Large-Scale Prompt Attacks on LLMs

Enhancing Adversarial Robustness of Vision Language Models via Adversarial Mixture Prompt Tuning

Energy-Driven Steering: Reducing False Refusals in Large Language Models

Strata-Sword: A Hierarchical Safety Evaluation towards LLMs based on Reasoning Complexity of Jailbreak Instructions

STAIR: Improving Safety Alignment with Introspective Reasoning

SeCon-RAG: A Two-Stage Semantic Filtering and Conflict-Free Framework for Trustworthy RAG