When Spectral Domain Meets Spatial Domain in Graph Neural Networks - Archive ouverte HAL Access content directly
Conference Papers Year :

When Spectral Domain Meets Spatial Domain in Graph Neural Networks

(1) , (1) , (1) , (1) , (1) , (1)
1

Abstract

Convolutional Graph Neural Networks (Con-vGNNs) are designed either in the spectral domain or in the spatial domain. In this paper, we provide a theoretical framework to analyze these neural networks, by deriving some equivalence of the graph convolution processes, regardless if they are designed in the spatial or the spectral domain. We demonstrate the relevance of the proposed framework by providing a spectral analysis of the most popular ConvGNNs (ChebNet, CayleyNet, GCN and Graph Attention Networks), which allows to explain their performance and shows their limits.
Fichier principal
Vignette du fichier
20.icml.gnn.pdf (933.37 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03088374 , version 1 (26-12-2020)

Identifiers

  • HAL Id : hal-03088374 , version 1

Cite

Balcilar Muhammet, Renton Guillaume, Héroux Pierre, Gaüzère Benoit, Adam Sébastien, et al.. When Spectral Domain Meets Spatial Domain in Graph Neural Networks. Thirty-seventh International Conference on Machine Learning (ICML 2020) - Workshop on Graph Representation Learning and Beyond (GRL+ 2020), Jul 2020, Vienna, Austria. ⟨hal-03088374⟩
218 View
127 Download

Share

Gmail Facebook Twitter LinkedIn More