Ming Zhang

4 papers North China Electric Power University

• Deeper with Riemannian Geometry: Overcoming Oversmoothing and Oversquashing for Graph Foundation Models
• Dynamic Bundling with Large Language Models for Zero-Shot Inference on Text-Attributed Graphs
• EvaLearn: Quantifying the Learning Capability and Efficiency of LLMs via Sequential Problem Solving
• SEGA: Shaping Semantic Geometry for Robust Hashing under Noisy Supervision