Reproduce “Toward Data-centric Directed Graph Learning: An Entropy-driven Approach” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xunkai Li, Zhengyu Wu, Kaichi Yu, Hongchao Qin, Guang Zeng, Rong-Hua Li, Guoren Wang
A data-centric directed graph learning framework called EDEN leverages entropy-driven knowledge distillation to improve graph neural network performance through hierarchical knowledge trees, achieving state-of-the-art results across 14 benchmark datasets while providing a flexible architecture that can enhance existing DiGNN models.
Key insights
- Combining topology-aware structural measurements with profile-aware node measurements enables better knowledge capture
- Node-adaptive knowledge distillation allows personalized transfer based on specific node requirements
- The framework can work both as a standalone DiGNN and as a plug-in module for existing models
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.