项目作者: gcucurull

项目描述 :
Graph Convolutional Networks in JAX
高级语言: Python
项目地址: git://github.com/gcucurull/jax-gcn.git
创建时间: 2020-04-02T14:24:26Z
项目社区:https://github.com/gcucurull/jax-gcn

开源协议:

下载


Graph Convolutional Networks in JAX

This repository implements GCNs in JAX (check it out on github). The code contains the model definition of a Graph Convolutional Network with two graph convolutional layers, following the model used in the paper Semi-Supervised Classification with Graph Convolutional Networks.

Usage

Run

python train.py

to train a model on the Cora dataset.

Good to know

I implemented a sparse matrix multiplication function to support sparse adjacency matrices, which is enabled by default. If you get any error with it, it can be disabled by adding the flag --no-sparse to the run command.

Cite

This is an implementation in JAX of the Semi-Supervised Classification with Graph Convolutional Networks paper. If you use it in your research, please cite the paper:

  1. @article{kipf2016semi,
  2. title={Semi-Supervised Classification with Graph Convolutional Networks},
  3. author={Kipf, Thomas N and Welling, Max},
  4. journal={arXiv preprint arXiv:1609.02907},
  5. year={2016}
  6. }