Structural Graph Probing of Vision-Language Models
（CVPR 2026）
When Do Hallucinations Arise？ A Graph Perspective on the Evolution of Path Reuse and Path Compression
（ ICML 2026） 从图结构的视角深入剖析了decoder-only Transformer架构的大型语言模型产生推理幻觉的内部机制。
Trap： Mitigating Poisoning-Based Backdoor Attacks by Treating Poison With Poison
IEEE Transactions on Dependable and Secure Computing（CCF A）
Finding the Correct Visual Evidence Without Forgetting： Mitigating Hallucination in LVLMs via Inter-Layer Visual Attention Discrepancy
（ICML 2026）