Junyu Liu

Medical-specific vision-language models surprisingly underutilize visual information in Japanese medical licensing exams, often performing well even when images are removed, highlighting a critical gap in their multimodal reasoning capabilities.

Quantum biosensors are evolving through four distinct generations, each leveraging progressively more exotic quantum phenomena to transcend classical limitations and enable adaptive inference directly within the quantum domain.