Pingchuan Ma

Enhancing CBMs by minimizing information leakage for better interpretability and trust.

DepthFM improves speed and accuracy in depth estimation using flow matching techniques.

A look into how MLLMs improve testing for image recognition systems.

This article explores using LLMs and simulations to enhance scientific research.

A new method creates stable 3D objects from images, maintaining appearance and structure.

A look into Normalizing Flows and their role in generative modeling.

A new approach enhances causal discovery accuracy in complex data relationships.

KANs bridge AI and science, simplifying complex problems and enhancing research.

Llama-AVSR merges audio and visual inputs for enhanced speech recognition accuracy.

A new method for stylizing 3D scenes enhances creativity in art and design.

New methods improve how AI describes images using language models.