Xingyuan Li

OmniPlan can cut network latency by nearly 98% while adapting to diverse user intents in real-time network planning.

Diffusion models can finally produce temporally stable video fusion by reframing the task as history-conditioned motion generation, sidestepping the limitations of optical flow and frame-by-frame processing.

Achieve sharper, more accurate infrared super-resolution in real-world conditions by disentangling thermal and structural degradations with a novel autoregressive framework.

Forget handcrafted losses: this paper uses human feedback and reinforcement learning to create infrared and visible image fusion that actually looks good to people.