Heterophilic Graph Learning

Tutorial on Learning On Graph (LoG) Conference 2024

Speakers: Sitao Luan, Chenqing Hua, QinCheng LU
Code Assistant: Jiaqi Zhu
Advisory Board: Guy Wolf, Xiao-Wen Chang.

Overview

This repository holds code and other relevant files about "Are Heterophily-Specific GNNs and Homophily Metrics Really Effective? Evaluation Pitfalls and New Benchmarks", which follows the tutorial presented at Learning on Graphs 2024. For additional details about the tutorial—including information about the speakers, slides, and video—please visit the official tutorial webpage

Colab Setup

All the content is designed to work seamlessly with Google Colab. Each Jupyter notebook includes installation commands for the necessary libraries at the beginning. At the top of each Jupyter notebook, you'll find an icon that lets you open it in Colab with just one click:

In this tutorial, we provide two notebooks:

Title	Link to Colab
Heterophilic-specific Models on Real World Datasets
Evaluation of Metrics on Synthetic Graphs

Citation

If you use the code or tutorial in your research, please cite as follows:

    @article{luan2024heterophily,
      title={Are Heterophily-Specific GNNs and Homophily Metrics Really Effective? Evaluation Pitfalls and New Benchmarks},
      author={Luan, Sitao and Lu, Qincheng and Hua, Chenqing and Wang, Xinyu and Zhu, Jiaqi and Chang, Xiao-Wen and Wolf, Guy and Tang, Jian},
      journal={arXiv preprint arXiv:2409.05755},
      year={2024}
    }

or

    @article{luan2024heterophilic,
      title={The heterophilic graph learning handbook: Benchmarks, models, theoretical analysis, applications and challenges},
      author={Luan, Sitao and Hua, Chenqing and Lu, Qincheng and Ma, Liheng and Wu, Lirong and Wang, Xinyu and Xu, Minkai and Chang, Xiao-Wen and Precup, Doina and Ying, Rex and others},
      journal={arXiv preprint arXiv:2407.09618},
      year={2024}
    }