Self-Attention Graph Pooling

Junhyun Lee, Inyeop Lee, Jaewoo Kang
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:3734-3743, 2019.

Abstract

Advanced methods of applying deep learning to structured data such as graphs have been proposed in recent years. In particular, studies have focused on generalizing convolutional neural networks to graph data, which includes redefining the convolution and the downsampling (pooling) operations for graphs. The method of generalizing the convolution operation to graphs has been proven to improve performance and is widely used. However, the method of applying downsampling to graphs is still difficult to perform and has room for improvement. In this paper, we propose a graph pooling method based on self-attention. Self-attention using graph convolution allows our pooling method to consider both node features and graph topology. To ensure a fair comparison, the same training procedures and model architectures were used for the existing pooling methods and our method. The experimental results demonstrate that our method achieves superior graph classification performance on the benchmark datasets using a reasonable number of parameters.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-lee19c, title = {Self-Attention Graph Pooling}, author = {Lee, Junhyun and Lee, Inyeop and Kang, Jaewoo}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {3734--3743}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/lee19c/lee19c.pdf}, url = {https://proceedings.mlr.press/v97/lee19c.html}, abstract = {Advanced methods of applying deep learning to structured data such as graphs have been proposed in recent years. In particular, studies have focused on generalizing convolutional neural networks to graph data, which includes redefining the convolution and the downsampling (pooling) operations for graphs. The method of generalizing the convolution operation to graphs has been proven to improve performance and is widely used. However, the method of applying downsampling to graphs is still difficult to perform and has room for improvement. In this paper, we propose a graph pooling method based on self-attention. Self-attention using graph convolution allows our pooling method to consider both node features and graph topology. To ensure a fair comparison, the same training procedures and model architectures were used for the existing pooling methods and our method. The experimental results demonstrate that our method achieves superior graph classification performance on the benchmark datasets using a reasonable number of parameters.} }
Endnote
%0 Conference Paper %T Self-Attention Graph Pooling %A Junhyun Lee %A Inyeop Lee %A Jaewoo Kang %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-lee19c %I PMLR %P 3734--3743 %U https://proceedings.mlr.press/v97/lee19c.html %V 97 %X Advanced methods of applying deep learning to structured data such as graphs have been proposed in recent years. In particular, studies have focused on generalizing convolutional neural networks to graph data, which includes redefining the convolution and the downsampling (pooling) operations for graphs. The method of generalizing the convolution operation to graphs has been proven to improve performance and is widely used. However, the method of applying downsampling to graphs is still difficult to perform and has room for improvement. In this paper, we propose a graph pooling method based on self-attention. Self-attention using graph convolution allows our pooling method to consider both node features and graph topology. To ensure a fair comparison, the same training procedures and model architectures were used for the existing pooling methods and our method. The experimental results demonstrate that our method achieves superior graph classification performance on the benchmark datasets using a reasonable number of parameters.
APA
Lee, J., Lee, I. & Kang, J.. (2019). Self-Attention Graph Pooling. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:3734-3743 Available from https://proceedings.mlr.press/v97/lee19c.html.

Related Material