Yuhang Zhang

PanoLOG's innovative partitioning strategy transforms panoramic 3D reconstruction, achieving top-tier rendering quality while slashing computational costs.

YouZhi-LLM achieves unprecedented concurrency and accuracy in financial LLMs by dramatically reducing KV-cache overhead, setting a new standard for deployment efficiency.

Achieve better image quality in Diffusion Transformers without extra training by dynamically trading off compute and cache risk.

By dynamically weighting RL updates based on the credibility of state estimates, this approach enables safer and more reliable autonomous navigation for USVs in uncertain environments.

Mimicking clinical diagnosis, this method boosts medical image classification by adaptively retrieving and reasoning over similar cases represented in multimodal knowledge graphs, leading to more accurate and explainable predictions.

Autonomous driving validation gets a shot in the arm: OVPD offers a proving ground dataset that fuses real vehicle dynamics with controllable virtual environments, enabling more realistic and diagnosable testing.

Forget domain adaptation tricks – GaussFly uses 3D Gaussian Splatting to create such photorealistic simulations that visuomotor policies trained on them transfer to the real world zero-shot.

Robots can now wipe surfaces more effectively using deformable objects, thanks to a new RL approach that simplifies state representation and efficiently processes contact feedback in simulation.

By actively exploring knowledge graphs with a differentiable neural-symbolic approach, NeuroSymActive achieves strong KGQA accuracy while drastically reducing the computational cost of graph lookups.