Bridging the Gap Between Spectral and Spatial Domains in Graph Neural Networks

This paper aims at revisiting Graph Convolutional Neural Networks by bridging the gap between spectral and spatial design of graph convolutions. We theoretically demonstrate some equivalence of the graph convolution process regardless it is designed in the spatial or the spectral domain... The obtained general framework allows to lead a spectral analysis of the most popular ConvGNNs, explaining their performance and showing their limits. Moreover, the proposed framework is used to design new convolutions in spectral domain with a custom frequency profile while applying them in the spatial domain. We also propose a generalization of the depthwise separable convolution framework for graph convolutional networks, what allows to decrease the total number of trainable parameters by keeping the capacity of the model. To the best of our knowledge, such a framework has never been used in the GNNs literature. Our proposals are evaluated on both transductive and inductive graph learning problems. Obtained results show the relevance of the proposed method and provide one of the first experimental evidence of transferability of spectral filter coefficients from one graph to another. Our source codes are publicly available at: https://github.com/balcilar/Spectral-Designed-Graph-Convolutions read more

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Node Classification CiteSeer with Public Split: fixed 20 nodes per class DSGCN Accuracy 73.3 # 12
Node Classification Cora: fixed 20 node per class DSGCN Accuracy 84.2 # 1
Node Classification Cora with Public Split: fixed 20 nodes per class DSGCN Accuracy 84.2% # 8
Graph Classification ENZYMES DSGCN-nodelabel Accuracy 65.13 # 9
Graph Classification ENZYMES DSGCN-allfeat Accuracy 78.39 # 1
Node Classification PPI DSGCN F1 99.09 # 8
Node Classification PubMed with Public Split: fixed 20 nodes per class DSGCN Accuracy 81.9% # 4

Methods