TY - JOUR
T1 - pathGCN
T2 - 39th International Conference on Machine Learning, ICML 2022
AU - Eliasof, Moshe
AU - Haber, Eldad
AU - Treister, Eran
N1 - Funding Information:
The research reported in this paper was supported by the Israel Innovation Authority through Avatar consortium. In addition, this work was supported in part by the Israeli Council for Higher Education (CHE) via the Data Science Research Center, Ben-Gurion University of the Negev, Israel. ME is supported by Kreitman High-tech scholarship.
Publisher Copyright:
Copyright © 2022 by the author(s)
PY - 2022/1/1
Y1 - 2022/1/1
N2 - Graph Convolutional Networks (GCNs), similarly to Convolutional Neural Networks (CNNs), are typically based on two main operations - spatial and point-wise convolutions. In the context of GCNs, differently from CNNs, a pre-determined spatial operator based on the graph Laplacian is often chosen, allowing only the point-wise operations to be learnt. However, learning a meaningful spatial operator is critical for developing more expressive GCNs for improved performance. In this paper we propose pathGCN, a novel approach to learn the spatial operator from random paths on the graph. We analyze the convergence of our method and its difference from existing GCNs. Furthermore, we discuss several options of combining our learnt spatial operator with point-wise convolutions. Our extensive experiments on numerous datasets suggest that by properly learning both the spatial and point-wise convolutions, phenomena like over-smoothing can be inherently avoided, and new state-of-the-art performance is achieved.
AB - Graph Convolutional Networks (GCNs), similarly to Convolutional Neural Networks (CNNs), are typically based on two main operations - spatial and point-wise convolutions. In the context of GCNs, differently from CNNs, a pre-determined spatial operator based on the graph Laplacian is often chosen, allowing only the point-wise operations to be learnt. However, learning a meaningful spatial operator is critical for developing more expressive GCNs for improved performance. In this paper we propose pathGCN, a novel approach to learn the spatial operator from random paths on the graph. We analyze the convergence of our method and its difference from existing GCNs. Furthermore, we discuss several options of combining our learnt spatial operator with point-wise convolutions. Our extensive experiments on numerous datasets suggest that by properly learning both the spatial and point-wise convolutions, phenomena like over-smoothing can be inherently avoided, and new state-of-the-art performance is achieved.
UR - http://www.scopus.com/inward/record.url?scp=85163105326&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85163105326
SN - 2640-3498
VL - 162
SP - 5878
EP - 5891
JO - Proceedings of Machine Learning Research
JF - Proceedings of Machine Learning Research
Y2 - 17 July 2022 through 23 July 2022
ER -