This is the code for Graph Neural Networks Exponentially Lose Expressive Power for Node Classification, accepted in International Conference on Learning Representations (ICLR) 2020.

• gnn_dynamics: code for experiments in Section 6.1
• gnn_normalization: code for experiment in Section 6.2

See README.md in each directory for details.

Citation

Oono, K., & Suzuki, T. (2020). Graph Neural Networks Exponentially Lose Expressive Power for Node Classification. In International Conference on Learning Representations, 2020. URL https://openreview.net/forum?id=S1ldO2EFPr.

@inproceedings{
oono2020graph,
title={Graph Neural Networks Exponentially Lose Expressive Power for Node Classification},
author={Kenta Oono and Taiji Suzuki},
booktitle={International Conference on Learning Representations},
year={2020},
url={https://openreview.net/forum?id=S1ldO2EFPr}
}

Acknowledgements

We thank Katsuhiko Ishiguro for providing a part of code for the experiments, Kohei Hayashi and Haru Negami Oono for giving us feedback and comments on the draft, and Keyulu Xu and anonymous reviewers for fruitful discussions via OpenReview.