This is the companion code for the paper: Spinelli I, Scardapane S, Uncini A, Adaptive Propagation Graph Convolutional Network, IEEE Transactions on Neural Networks and Learning Systems, 2020.
We introduce the adaptive propagation graph convolutional network (AP-GCN), a variation of GCN wherein each node selects automatically the number of propagation steps performed across the graph.
All the code for the models described in the paper can be found in model.py. An example of use which can be quickly extendet to the full experimental evaluation is provided in AP-GCN_demo.ipynb.
[1] Kipf, T.N. and Welling, M., 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907.
[2] Klicpera J., Bojchevski A., Günnemann S. Predict then Propagate: Graph Neural Networks meet Personalized PageRank. arXiv preprint arXiv:1810.05997
[3] Fey M., Lenssen J.E. Fast Graph Representation Learning with PyTorch Geometric. arXiv preprint arXiv:1903.02428
Many thanks to the authors of [2] for making their code public. Our repo is based on their work for a fair comparison between algorithms. Many thanks also to the maintainers [3] for such an awesome open-source library.
Please cite our paper if you use this code in your own work:
@ARTICLE{spinelli2020apgcn,
author={I. {Spinelli} and S. {Scardapane} and A. {Uncini}},
journal={IEEE Transactions on Neural Networks and Learning Systems},
title={Adaptive Propagation Graph Convolutional Network},
year={2020},
volume={},
number={},
pages={1-6},}