Algorithms for object categorization have demonstrated great achievements in learning, processing and comparing objects. The problem occurs if the database of solved objects is too large and the classification time is being increased proportionally. We present a solution by optimalization of obtained knowledge from learning, using the clustering of a modified neural network. Several methods for modifying a number of clusters to increase the speed or the accuracy of classification are shown. Additionally, the proposed algorithm is able to compute the similarity between clusters of various classes and then create statements about the homogeneity, independence, inferiority and similarity of classes. A series of experiments is shown, some of them also as a part of the multi-agent system, Nao robots and gestures and emotions. The analysis could be useful for students and researchers of the computational intelligence or anyone else who may be considering utlizing neural networks for the object categorization and analysis of obtained knowledge from images or from a video stream.
Les informations fournies dans la section « Synopsis » peuvent faire référence à une autre édition de ce titre.
Algorithms for object categorization have demonstrated great achievements in learning, processing and comparing objects. The problem occurs if the database of solved objects is too large and the classification time is being increased proportionally. We present a solution by optimalization of obtained knowledge from learning, using the clustering of a modified neural network. Several methods for modifying a number of clusters to increase the speed or the accuracy of classification are shown. Additionally, the proposed algorithm is able to compute the similarity between clusters of various classes and then create statements about the homogeneity, independence, inferiority and similarity of classes. A series of experiments is shown, some of them also as a part of the multi-agent system, Nao robots and gestures and emotions. The analysis could be useful for students and researchers of the computational intelligence or anyone else who may be considering utlizing neural networks for the object categorization and analysis of obtained knowledge from images or from a video stream.
Peter Smolár,PhD: studied branch of Artificial Intelligence at the Technical University of Kosice, Faculty of Electrical Engineering and Informatics, Department of Cybernetics and Artificial Intelligence.His research includes computational intelligence, particularly neural networks and fuzzy systems, humanoid robotics and object recognition.
Les informations fournies dans la section « A propos du livre » peuvent faire référence à une autre édition de ce titre.
EUR 9,70 expédition depuis Allemagne vers France
Destinations, frais et délaisVendeur : moluna, Greven, Allemagne
Kartoniert / Broschiert. Etat : New. N° de réf. du vendeur 5142870
Quantité disponible : Plus de 20 disponibles
Vendeur : Revaluation Books, Exeter, Royaume-Uni
Paperback. Etat : Brand New. 188 pages. 8.66x5.91x0.43 inches. In Stock. N° de réf. du vendeur __3659247952
Quantité disponible : 1 disponible(s)
Vendeur : Revaluation Books, Exeter, Royaume-Uni
Paperback. Etat : Brand New. 188 pages. 8.66x5.91x0.43 inches. In Stock. N° de réf. du vendeur 3659247952
Quantité disponible : 1 disponible(s)
Vendeur : Mispah books, Redhill, SURRE, Royaume-Uni
paperback. Etat : New. New. book. N° de réf. du vendeur ERICA82936592479526
Quantité disponible : 1 disponible(s)