Relational Pooling for Graph Representations

Ryan Murphy, Balasubramaniam Srinivasan, Vinayak Rao, Bruno Ribeiro
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:4663-4673, 2019.

Abstract

This work generalizes graph neural networks (GNNs) beyond those based on the Weisfeiler-Lehman (WL) algorithm, graph Laplacians, and diffusions. Our approach, denoted Relational Pooling (RP), draws from the theory of finite partial exchangeability to provide a framework with maximal representation power for graphs. RP can work with existing graph representation models and, somewhat counterintuitively, can make them even more powerful than the original WL isomorphism test. Additionally, RP allows architectures like Recurrent Neural Networks and Convolutional Neural Networks to be used in a theoretically sound approach for graph classification. We demonstrate improved performance of RP-based graph representations over state-of-the-art methods on a number of tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-murphy19a, title = {Relational Pooling for Graph Representations}, author = {Murphy, Ryan and Srinivasan, Balasubramaniam and Rao, Vinayak and Ribeiro, Bruno}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {4663--4673}, 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/murphy19a/murphy19a.pdf}, url = {https://proceedings.mlr.press/v97/murphy19a.html}, abstract = {This work generalizes graph neural networks (GNNs) beyond those based on the Weisfeiler-Lehman (WL) algorithm, graph Laplacians, and diffusions. Our approach, denoted Relational Pooling (RP), draws from the theory of finite partial exchangeability to provide a framework with maximal representation power for graphs. RP can work with existing graph representation models and, somewhat counterintuitively, can make them even more powerful than the original WL isomorphism test. Additionally, RP allows architectures like Recurrent Neural Networks and Convolutional Neural Networks to be used in a theoretically sound approach for graph classification. We demonstrate improved performance of RP-based graph representations over state-of-the-art methods on a number of tasks.} }
Endnote
%0 Conference Paper %T Relational Pooling for Graph Representations %A Ryan Murphy %A Balasubramaniam Srinivasan %A Vinayak Rao %A Bruno Ribeiro %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-murphy19a %I PMLR %P 4663--4673 %U https://proceedings.mlr.press/v97/murphy19a.html %V 97 %X This work generalizes graph neural networks (GNNs) beyond those based on the Weisfeiler-Lehman (WL) algorithm, graph Laplacians, and diffusions. Our approach, denoted Relational Pooling (RP), draws from the theory of finite partial exchangeability to provide a framework with maximal representation power for graphs. RP can work with existing graph representation models and, somewhat counterintuitively, can make them even more powerful than the original WL isomorphism test. Additionally, RP allows architectures like Recurrent Neural Networks and Convolutional Neural Networks to be used in a theoretically sound approach for graph classification. We demonstrate improved performance of RP-based graph representations over state-of-the-art methods on a number of tasks.
APA
Murphy, R., Srinivasan, B., Rao, V. & Ribeiro, B.. (2019). Relational Pooling for Graph Representations. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:4663-4673 Available from https://proceedings.mlr.press/v97/murphy19a.html.

Related Material