floregol / bgcn_copying

Bayesian Graph Convolutional Network with Copying

This repository contains the code from the paper :

S. Pal, F. Regol, and M.J. Coates, “Bayesian graph convolutional neural networks using node copying,” in Proc. Workshop on Learning and Reasoning with Graph-Structured Data at Int. Conf. Machine Learning, June 2019.

The included datasets are : Cora, Citeseer and Pubmed.

REQUIREMENTS AND INSTALLATION

python 3
tensorflow
networkx
numpy
scipy
scikit_learn

You can install the required packages using the provided requirement file

pip install -f requirements.txt

USAGE

To run the program with default parameters:

python3 Run_trials.py

CONTACT

[email protected]

Cite

@inproceedings{pal2019,
title={Bayesian graph convolutional neural networks using node copying},
author={S. Pal, F. Regol, and M.J. Coates},
booktitle={Proc. Workshop on Learning and Reasoning with Graph-Structured Data at Int. Conf. Machine Learning},
year={2019}
}