Maximally Expressive GNNs for Outerplanar Graphs

Code for our paper Maximally Expressive GNNs for Outerplanar Graphs (TMLR 2024). Previous versions appeared at GLF@NeurIPS (2023) and LoG (Extended Abstract, 2023).

Setup

Clone this repository and open the directory

Add this directory to the python path. Let $PATH be the path to where this repository is stored (i.e. the result of running pwd).

export PYTHONPATH=$PYTHONPATH:$PATH

Create a conda environment (this assume miniconda is installed)

conda create --name GNNs

Activate environment

conda activate GNNs

Install dependencies

conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 -c pytorch
conda install -c pyg pyg=2.2.0
python -m pip install -r requirements.txt

Replicating the experiments

Results can be found in the results directory.

Baselines:

bash Scripts/experiments_GIN_baselines.sh
bash Scripts/experiments_GCN_baselines.sh
bash Scripts/experiments_GAT_baselines.sh

CAT models:

bash Scripts/experiments_GIN_cat.sh
bash Scripts/experiments_GCN_cat.sh
bash Scripts/experiments_GAT_cat.sh

Benchmark GIN vs CAT+GIN runtime:

bash Scripts/benchmark_training.sh 

Benchmark CAT pre-processing time (results in terminal):

python Scripts/benchmark_cat.py

Compute directed effective resistance for CAT:

python Exp/resistance.py

Citation

Please cite us as

@inproceedings{Outerplanar-GNNs-GLF,
  title={Maximally Expressive {GNNs} for Outerplanar Graphs},
  author={Bause, Franka and Jogl, Fabian and Indri, Patrick and Drucks, Tamara and Kriege, Nils Morten and Gärtner, Thomas and Welke, Pascal and Thiessen, Maximilian},
  booktitle={TMLR},
  year={2024}
}