• Stars
    star
    287
  • Rank 143,384 (Top 3 %)
  • Language
    C++
  • License
    MIT License
  • Created over 8 years ago
  • Updated over 3 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

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

Docker

Dockerfile contains all the required installations (including Intel MKL and TBB) above. Only additional requirement is to provide NVIDIA*.run script that will load the same NVIDIA driver of host into the target. Then to build the container, execute:

docker build -t "graphnn:test" .

To run it:

docker run --runtime=nvidia graphnn:test bash

If above command fails for a reason, refer to https://github.com/NVIDIA/nvidia-docker. If no error occurs, you can simply follow the below instructions and execute them in the container without failure.

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}
}