Lifeng Sun

Medical VLMs can achieve state-of-the-art visual reasoning without any human annotations, thanks to a novel reinforcement learning framework that leverages model uncertainty and rollout agreement.

SubFLOT tackles federated learning's heterogeneity problem by cleverly using optimal transport to create personalized submodels on the server, sidestepping the computational burden of client-side pruning.

Image watermarks can now withstand a barrage of adversarial, distortion, and regeneration attacks, thanks to a decoupled training strategy that boosts robustness by up to 46% without sacrificing image quality.

LLMs can now guide video streaming optimization, outperforming traditional saliency models and human annotation in predicting content importance for both VOD and live streams.