Accéder directement au contenu Accéder directement à la navigation
Pré-publication, Document de travail

Graph Kernels Based on Linear Patterns: Theoretical and Experimental Comparisons

Linlin Jia 1 Benoit Gaüzère 1 Paul Honeine 1
1 DocApp - LITIS - Equipe Apprentissage
LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
Abstract : Graph kernels are powerful tools to bridge the gap between machine learning and data encoded as graphs. Most graph kernels are based on the decomposition of graphs into a set of patterns. The similarity between two graphs is then deduced from the similarity between corresponding patterns. Kernels based on linear patterns constitute a good trade-off between accuracy performance and computational complexity. In this work, we propose a thorough investigation and comparison of graph kernels based on different linear patterns, namely walks and paths. First, all these kernels are explored in detail, including their mathematical foundations, structures of patterns and computational complexity. Then, experiments are performed on various benchmark datasets exhibiting different types of graphs, including labeled and unlabeled graphs, graphs with different numbers of vertices, graphs with different average vertex degrees, cyclic and acyclic graphs. Finally, for regression and classification tasks, performance and computational complexity of kernels are compared and analyzed, and suggestions are proposed to choose kernels according to the types of graph datasets. This work leads to a clear comparison of strengths and weaknesses of these kernels. An open-source Python library containing an implementation of all discussed kernels is publicly available on GitHub to the community, thus allowing to promote and facilitate the use of graph kernels in machine learning problems.
Liste complète des métadonnées

Littérature citée [35 références]  Voir  Masquer  Télécharger

https://hal-normandie-univ.archives-ouvertes.fr/hal-02053946
Contributeur : Paul Honeine <>
Soumis le : vendredi 1 mars 2019 - 15:56:20
Dernière modification le : vendredi 15 mai 2020 - 12:00:03
Document(s) archivé(s) le : jeudi 30 mai 2019 - 15:54:56

Fichier

2019:03:01 Graph_Kernels_on_Li...
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-02053946, version 1

Citation

Linlin Jia, Benoit Gaüzère, Paul Honeine. Graph Kernels Based on Linear Patterns: Theoretical and Experimental Comparisons. 2019. ⟨hal-02053946⟩

Partager

Métriques

Consultations de la notice

347

Téléchargements de fichiers

251