missingboyzfy / GraphHop

GraphHop: An Enhanced Label PropagationMethod for Node Classification

This respository contains the PyTorch implementation of GraphHop for the task of semi-supervised classification of nodes in a graph, as described in our paper:

Tian Xie, Bin Wang, C.-C. Jay Kuo, GraphHop: An Enhanced Label PropagationMethod for Node Classification. [paper]

Dependencies

• torch == 1.5.0
• numpy == 1.18.1
• scipy == 1.4.1
• pytorch geometric

RUN

sh run_model.sh

You may change the hyperparameters inside the shell script.

Citation

If you are use this code for your research, please cite our paper.

@misc{xie2021graphhop,
      title={GraphHop: An Enhanced Label Propagation Method for Node Classification}, 
      author={Tian Xie and Bin Wang and C. -C. Jay Kuo},
      year={2021},
      eprint={2101.02326},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}