Jionglong Su

Simulating adversarial debate between specialized agents dramatically reduces hallucinations in medical diagnosis MLLMs, surpassing single-agent baselines in accuracy and trustworthiness.

Achieve state-of-the-art pulmonary nodule segmentation by distilling language guidance into a graph-reasoning framework, all while fine-tuning less than 1% of the parameters.

Forget brute-force scaling: smart data curation and reward-guided optimization can dramatically boost Chinese-to-Southeast Asian language translation, leaving larger models in the dust.