Community Detection with Deep Graph Neural Networks / Imad Eddine Kettaf  

Public ISBD Titre : Community Detection with Deep Graph Neural Networks Type de document : texte imprimé Auteurs : Imad Eddine Kettaf, Auteur ; Abdelouahab Moussaoui, Directeur de thèse Année de publication : 2022 Importance : 1 vol (79 f .) Format : 29cm Langues : Français (fre) Catégories : Thèses & Mémoires:Informatique Mots-clés : Informatique Index. décimale : 004 Informatique Résumé : n recent years, Deep Neural Networks (DNN) have achieved significant advances in image categorization and speech recognition. Using DNN for community identification, such as in social networks, remains a difficult challenge. Furthermore, the latest sophisticated techniques of executing community detection tasks employ Graph Neural Networks (GNN), a novel DNN methodology. In this thesis, we investigated the applicability of these new techniques for detecting communities by creating a new GNN-based learning model that combines two of the most recent state-of-the-art approaches in community detection, Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT). Our suggested model is made up of four layers and performs semi-supervised node classification to discover communities. The first two layers are GCN layers, followed by a GAT layer, then ended with another GCN layer. We utilized four distinct datasets from the pytorch-geometric package for training and testing our model. These datasets are ”Cora,” ”Citeseer,” ”Pubmed,” and ”Ego-Facebook.” This architecutre of our model enabled it to outperform the other comparative techniques as well as certain state-of-the-art approaches. Côte titre : MAI/0607 En ligne : https://drive.google.com/file/d/1bo_KtDR87-Ef9Z47Q7vUuWqfQnQmG0Q9/view?usp=share [...] Format de la ressource électronique : pdf Community Detection with Deep Graph Neural Networks [texte imprimé] / Imad Eddine Kettaf, Auteur ; Abdelouahab Moussaoui, Directeur de thèse . - 2022 . - 1 vol (79 f .) ; 29cm. Langues : Français (fre) Catégories : Thèses & Mémoires:Informatique Mots-clés : Informatique Index. décimale : 004 Informatique Résumé : n recent years, Deep Neural Networks (DNN) have achieved significant advances in image categorization and speech recognition. Using DNN for community identification, such as in social networks, remains a difficult challenge. Furthermore, the latest sophisticated techniques of executing community detection tasks employ Graph Neural Networks (GNN), a novel DNN methodology. In this thesis, we investigated the applicability of these new techniques for detecting communities by creating a new GNN-based learning model that combines two of the most recent state-of-the-art approaches in community detection, Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT). Our suggested model is made up of four layers and performs semi-supervised node classification to discover communities. The first two layers are GCN layers, followed by a GAT layer, then ended with another GCN layer. We utilized four distinct datasets from the pytorch-geometric package for training and testing our model. These datasets are ”Cora,” ”Citeseer,” ”Pubmed,” and ”Ego-Facebook.” This architecutre of our model enabled it to outperform the other comparative techniques as well as certain state-of-the-art approaches. Côte titre : MAI/0607 En ligne : https://drive.google.com/file/d/1bo_KtDR87-Ef9Z47Q7vUuWqfQnQmG0Q9/view?usp=share [...] Format de la ressource électronique : pdf

Exemplaires (1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
MAI/0607	MAI/0607	Mémoire	Bibliothéque des sciences	Anglais	Disponible Disponible

---

1 (1 - 1 / 1)