Node-wise Diffusion for Scalable Graph Learning

This repository contains the codes of NIGCN, data process, and split generation codes. For detailed information about NIGCN, please refer to the paper https://arxiv.org/abs/2305.14000. If any issues are observed, please contact Keke Huang, thanks!

Requirements

• CUDA 10.1.243
• python 3.6.10
• pytorch 1.4.0
• GCC 5.4.0
• cnpy
• swig-4.0.1

Datasets

The tested datasets can be downloaded from the corresponding citation in the papers.

In particular, folder preprocessing contains the codes for data processing and split generation.

Compilation

make

Running the code

Folder run_scripts contains the script to run NIGCN for each datasets.

Citation

Please cite our paper if it is relevant to your work, thanks!

@inproceedings{Huang0LYX23,
  author       = {Keke Huang and
                  Jing Tang and
                  Juncheng Liu and
                  Renchi Yang and
                  Xiaokui Xiao},
  title        = {Node-wise Diffusion for Scalable Graph Learning},
  booktitle    = {{WWW}},
  pages        = {1723--1733},
  year         = {2023},
}