Veuillez utiliser cette adresse pour citer ce document :
https://zone.biblio.laurentian.ca/handle/10219/3503
Titre: | A neuroevolutionary neural network-based collaborative filtering recommendation system |
Auteurs: | Monemian, Seyedamin |
Mots clés: | NeuroEvolution,;neural network;neuroevolution of augmenting;topologies (NEAT);recommendation systems |
Date publié: | 29-mai-2020 |
Abstrait: | The success of a neural network-based recommendation system depends on finding an architecture to fit the task. The Fixed Topology NeuroEvolution approach is considered outdated when compared with an automated method for optimizing neural network structures. A NeuroEvolutionary algorithm has been developed in order to enhance the structure of a neural network to yield a more accurate recommendation system based on some of the existing benchmarks for measuring recommendation capability. The genetic algorithm ensures us to gain a more efficient topology produced by different generations through each step of the evolution process. Results show that this method performs better than many other algorithms and is close to the Singular Value Decomposition based algorithm developed by Funk as a benchmark standard. |
URI: | https://zone.biblio.laurentian.ca/handle/10219/3503 |
Apparaît dans les collections: | Computational Sciences - Master's theses Master's Theses |
Fichiers dans cet item:
Fichier | Description | Taille | Format | |
---|---|---|---|---|
S Monemian Thesis Final.pdf | 1.03 MB | Adobe PDF | Parcourir/Ouvrir |
Tous les documents dans DSpace sont protégés par copyright, avec tous droits réservés.