Huimin Zeng

LLMs can now generate more relevant and factual movie recommendations by dynamically bridging retrieval and generation with a novel reinforcement learning approach.

Achieve real-time (76 FPS) HDR novel view synthesis with significantly improved reconstruction of high dynamic range details by explicitly modeling scene reflectance and illumination.

MLLMs that ace standard Referring Expression Comprehension benchmarks still stumble when faced with images designed to eliminate shortcuts, revealing a surprising lack of robust visual reasoning.

LLMs could be the key to unlocking unified hydrological models by bridging data silos, synthesizing knowledge, and enabling modular, reasoning-driven development.