alexbouril / graphnn

Training computational graph on top of structured data (string, graph, etc)

Dec. 22, 2017 update: pytorch version of structure2vec

For people who prefer python, here is the pytorch implementation of s2v:

https://github.com/Hanjun-Dai/pytorch_structure2vec

graphnn

Document

(Doxygen) http://www.cc.gatech.edu/~hdai8/graphnn/html/annotated.html

Prerequisites

Tested under Ubuntu 14.04, 16.04 and Mac OSX 10.12.6

Download and install cuda from https://developer.nvidia.com/cuda-toolkit

wget http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1404/x86_64/cuda-repo-ubuntu1404_8.0.44-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1404_8.0.44-1_amd64.deb
sudo apt-get update
sudo apt-get install cuda

in .bashrc, add the following path (suppose you installed to the default path)

export CUDA_HOME=/usr/local/cuda
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH

Download and install intel mkl

in .bashrc, add the following path

source {path_to_your_intel_root/name_of_parallel_tool_box}/bin/psxevars.sh

Build static library

cp make_common.example make_common
modify configurations in make_common file
make -j8

Run example

Run mnist

cd examples/mnist
make
./run.sh

Run graph classification

cd examples/graph_classification
make
./local_run.sh

The 5 datasets under the data/ folder are commonly used in graph kernel. 

Reference

@article{dai2016discriminative,
  title={Discriminative Embeddings of Latent Variable Models for Structured Data},
  author={Dai, Hanjun and Dai, Bo and Song, Le},
  journal={arXiv preprint arXiv:1603.05629},
  year={2016}
}