Gao Huang

A new approach emphasizes reward understanding for better RL decision insights.

GORL enhances offline RL by leveraging expert data for better decision-making.

Deformable attention improves efficiency and accuracy in visual recognition tasks.

A new text encoder tackles challenges in visual text rendering.

New model improves image processing in multimodal systems.

DiveR-CT improves automated red teaming for better safety assessments.

A new method to enhance the safety of language models with less effort.

A study on techniques for recovering low-rank Toeplitz matrices from limited measurements.

New dataset and model improve head detection in crowded scenes.

A new framework enhances image generation speed and quality in diffusion transformers.

A new method improves cardiac ultrasound guidance through historical data.

SMART enhances open-vocabulary segmentation by improving mask classification techniques.

A new framework makes robots smarter and more efficient for everyday tasks.

A new solver improves generalization in physics-informed neural networks.