Skip to Main content Skip to Navigation
Conference papers

When Spectral Domain Meets Spatial Domain in Graph Neural Networks

Balcilar Muhammet 1 Renton Guillaume 1 Héroux Pierre 1 Gaüzère Benoit 1 Adam Sébastien 1 Paul Honeine 1 
1 DocApp - LITIS - Equipe Apprentissage
LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
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.
Complete list of metadata
Contributor : Paul Honeine Connect in order to contact the contributor
Submitted on : Saturday, December 26, 2020 - 10:57:12 AM
Last modification on : Wednesday, March 2, 2022 - 10:10:12 AM
Long-term archiving on: : Monday, March 29, 2021 - 4:42:50 PM


Files produced by the author(s)


  • HAL Id : hal-03088374, version 1


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⟩



Record views


Files downloads