Wenjie Wang

Chart Deep Research in LVLMs via Parallel Relative Policy Optimization

STEDiff: Revealing the Spatial and Temporal Redundancy of Backdoor Attacks in Text-to-Image Diffusion Models

Safeguarding Multimodal Knowledge Copyright in the RAG-as-a-Service Environment

CAP: Improving the Robustness of LLM-as-a-Judge Against Adversarial Score Manipulation via Comparative Augmented Prompting

Auto-Search and Refinement: An Automated Framework for Gender Bias Mitigation in Large Language Models

Don’t Say No: Jailbreaking LLM by Suppressing Refusal

Certified PEFTSmoothing: Parameter-Efficient Fine-Tuning with Randomized Smoothing