Chu-Song Chen

RECON: Multimodal GraphRAG for Visually Rich Documents with Intra-Page Reflection and Inter-Page Connection

Story2Screen: Multimodal Story Customization for Long Consistent Visual Sequences

Safety Depth in Large Language Models: A Markov Chain Perspective

Information-Theoretical Principled Trade-off between Jailbreakability and Stealthiness on Vision Language Models