一作NDSS 2026Oral· Network & Distributed System SecurityQ-MLLM: Vector Quantization for Robust Multimodal Large Language Model SecurityWei Zhao · Zhe Li · Yige Li · Jun Sun安全多模态效率与系统OpenReview
作者位次 #2/4ICLR 2026Poster· Social Aspects / Fairness, Equity, Justice and SafetyWhere Did It Go Wrong? Attributing Undesirable LLM Behaviors via Representation Gradient TracingZhe Li · Wei Zhao · Yige Li · Jun Sun社会与人本安全可解释性OpenReview