G3NN

This repo provides a pytorch implementation for the 4 instantiations of the flexible generative framework as described in the following paper:

A Flexible Generative Framework for Graph-based Semi-supervised Learning

Jiaqi Ma*, Weijing Tang*, Ji Zhu, and Qiaozhu Mei. NeurIPS 2019.

(*: equal contribution)

Requirements

See environment.yml. Run conda torch_env create -f environment.yml to install the required packages.

Run the code

Example: python main.py --model lsm_gcn --dataset cora

Cite

@inproceedings{ma2019flexible,
  title={A Flexible Generative Framework for Graph-based Semi-supervised Learning},
  author={Ma, Jiaqi and Tang, Weijing and Zhu, Ji and Mei, Qiaozhu},
  booktitle={Advances in Neural Information Processing Systems},
  pages={3276--3285},
  year={2019}
}