Motifs locaux et super-graphe pour la classification de graphes symboliques avec des réseaux convolutionnels

Évariste Daller 1 Luc Brun 1 Sébastien Bougleux 1 Olivier Lézoray 1
1 Equipe Image - Laboratoire GREYC - UMR6072
GREYC - Groupe de Recherche en Informatique, Image, Automatique et Instrumentation de Caen
Abstract : Convolutional neural networks (CNN) have deeply impacted the field of machine learning. These networks designed to process objects with a fixed topology readily apply to images, videos and sounds but can not be easily extended to structures with an arbitrary topology such as graphs. Examples of applications of machine learning to graphs include the prediction of the properties of molecular graphs or the classification of 3D meshes. Within the symbolic graphs framework, we propose a method to extend networks based on a fixed topology to input graphs with an arbitrary topology. We also propose an enriched feature vector attached to each vertex of a chemical graph in order to improve the prediction of its properties as well as a new bottleneck layer allowing to connect arbitrary topological graphs on a fully connected layer.
Liste complète des métadonnées

Cited literature [19 references]  Display  Hide  Download

https://hal-normandie-univ.archives-ouvertes.fr/hal-01796587
Contributor : Évariste Daller <>
Submitted on : Monday, May 21, 2018 - 10:32:48 AM
Last modification on : Thursday, February 7, 2019 - 5:45:52 PM
Document(s) archivé(s) le : Tuesday, September 25, 2018 - 10:39:03 AM

File

rfiap-2018.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01796587, version 1

Citation

Évariste Daller, Luc Brun, Sébastien Bougleux, Olivier Lézoray. Motifs locaux et super-graphe pour la classification de graphes symboliques avec des réseaux convolutionnels. Reconnaissance des Formes, Image, Apprentissage et Perception, RFIAP, AFRIF (Association Française pour la Reconnaissance et l'Interprétation des Formes), Jun 2018, Marne-la-Vallée, France. ⟨hal-01796587⟩

Share

Metrics

Record views

109

Files downloads

142