Yong Xia

Stop your Med-VQA model from "hallucinating" answers: this plug-in framework forces LLMs to actually *look* at the image by bottlenecking visual information through question-conditioned cues.

Regularizing Lipschitz constants in MLPs within neural oscillators provably and practically enhances generalization, offering a path to more robust learning of complex dynamical systems.

RL for radiology report generation can achieve state-of-the-art results with significantly less data by prioritizing data quality and weighting tokens based on their diagnostic importance.