Multiscale Graph Neural Networks in PyTorch

PyTorch implementation of Community Detection with Graph Neural Networks [1].

For a high-level introduction to Graph Neural Networks, see:

Thomas Kipf, Graph Convolutional Networks (2016)

Note: This code is based on the Lua implementation in https://github.com/joanbruna/GNN_community . This re-implementation serves as a proof of concept and is not intended for reproduction of the results reported in [1].

The experiments on real-world community detection are based on the Snap graphs with ground-truth community (Stanford Network Analysis Project) [2].

Installation

python setup.py install

Requirements

• PyTorch 0.2.0
• Python 3.6

Usage

cd multiscalegnn

python snap.py --graph 'dblp' --data_dir './../data/'

References

[1]J. Bruna and L. Li, Community Detection with Graph Neural Networks, 2017.

[2]Stanford Network Analysis Project.