Tao Hu

Local electronic observables can now be evaluated without full orbital diagonalization by structuring reusable Fourier probes into a spatial hierarchy that achieves near-linear scaling across both insulators and metals.

CLIP's cross-modal gap makes text-derived classifier weights actively harm class-incremental learning—discarding the text branch entirely in favor of closed-form visual SVMs delivers superior continual recognition.

Multimodal continual learning does not require rigid parameter isolation or massive replay buffers—external reasoning retrieval can dynamically compose modular capability paths while subspace recycling safely prevents cross-task interference.

EyeControl transforms image retouching by enabling users to enhance visual focus with just a few clicks, achieving expert-level results without requiring professional skills.

Real-time risk assessment in autonomous driving just got a major upgrade, achieving state-of-the-art performance with a novel framework that combines perception and structured reasoning.