This is a reproduction of Identity-Aware Graph Neural Networks by Zhihao Dong and Jeremy Banks.

@report{You2021,
   author = {Jiaxuan You and Jonathan M Gomes-Selman and Rex Ying and Jure Leskovec},
   keywords = {Data Mining & Knowledge Management: Graph Mining & Social Network Analysis & Community,Machine Learning: Graph-based Machine Learning,Machine Learning: Relational Learning,Machine Learning: Representation Learning},
   title = {Identity-aware Graph Neural Networks},
   url = {http://snap.stanford.edu/idgnn},
   year = {2021},
}

To install requirements:

Requirements

• CPU or NVIDIA GPU, Linux, Python3
• PyTorch, various Python packages; Instructions for installing these dependencies are found below

1. Python environment (Optional): We recommend using Conda package manager

conda create -n graphgym python=3.7
source activate graphgym

2. Pytorch: Install PyTorch. We have verified GraphGym under PyTorch 1.8.0, and GraphGym should work with PyTorch 1.4.0+. For example:

# CUDA versions: cpu, cu92, cu101, cu102, cu101, cu111
pip install torch==1.8.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

3. Pytorch Geometric: Install PyTorch Geometric, follow their instructions. For example:

# CUDA versions: cpu, cu92, cu101, cu102, cu101, cu111
# TORCH versions: 1.4.0, 1.5.0, 1.6.0, 1.7.0, 1.8.0
CUDA=cu101
TORCH=1.8.0
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-geometric

4. GraphGym and other dependencies:

git clone https://github.com/snap-stanford/GraphGym
cd GraphGym
pip install -r requirements.txt
pip install -e .  # From latest verion
pip install graphgym # (Optional) From pypi stable version

To train the model(s) in the paper, run this command:

You can run the training by selecting the model using the following formula:

BASE=0
FAST=1
FULL=2
BASE_MODEL='gcn' # or any of the following: ['sage', 'gat', 'gin']
AUGMENT=$BASE # or any of the following: [FAST, FULL]
declare -a models=("${BASE_MODEL}conv_tf"
                   "${BASE_MODEL}convFast_tf"
                   "id${BASE_MODEL}_tf")
MODEL=${models[$AUGMENT]}

python main_zd.py --model $MODEL

You can also simply type the following, using the available models:

python main_zd.py --model ginconv_tf

The evaluations of datasets are automatically placed in the results folder after training. Simply navigate to the results/val/final folder and locate the file associated with the model that you trained.

Our model achieves the following performance on the Node classification task using the datasets from the original paper:

We fully acknowledge that the results on the real-world datasets with greater than 90% accuracy are likely erroneous, and we ask you to bear with us while we investigate this issue.

We did not find a licence section on the page of the original authors, but they designed the GraphGym framework for the express purpose of allowing contributions. I freely welcome you forking and continuing any of our work, however I am not willing to accept pull requests. If you fork and continue this work, please let me know and I will place a link to your fork in this readme. The original authors strongly encourage people submitting pull requests to contribute to their project. I encourage you to contribute any PyTorch modules to the original GitHub repo available here: https://github.com/snap-stanford/GraphGym