Optimal structure for large-scale data clustering based on support vector machine and fuzzy rules - Archive ouverte HAL
Pré-Publication, Document De Travail Année : 2022

Optimal structure for large-scale data clustering based on support vector machine and fuzzy rules

Résumé

Data clustering is a method for classifying similar data that used in various sciences for many years and many algorithms designed in this field. Recent clustering research has led to hybrid methods that are more robust and accurate. Combined clustering first tries to produce primary clustering that is as scattered as possible and then combines the results by applying an agreement function. In this research, a combination of fuzzy clustering and support vector machine used for classification. The hybrid network (FS-FCSVM) is an efficient fuzzy clustering operation performed on the input data. The network parameters trained with SVM, achieves a network with high generalizability. The number of rules in such systems is smaller than fuzzy systems and cussed a lower computation time. Data clustering is a technique for identifying related data that has been utilized for many years in a variety of fields. Numerous algorithms have been developed in this area. Recent advances in clustering research have produced hybrid techniques that are more reliable and precise. In order to combine the findings, combined clustering first attempts to create primary clustering that is as dispersed as feasible. In this study, categorization was done using a hybrid of fuzzy clustering and support vector machines. The input data are effectively fuzzy clustered using the hybrid network (FS-FCSVM). With SVM-trained network parameters, a highly generalizable network is produced. These systems have fewer rules than fuzzy systems and need less computing time. In this study, the reduction clustering method used before fuzzy clustering. The main idea of reduction clustering is to search for high-density regions in the data space characteristic. Each point that has the largest number of neighbors selected as the center of the cluster. In other words, the reduction clustering technique used to select feature points that are more different and less similar to other points. In this paper, the idea is to use differential clustering to find the exact center points of
Fichier principal
Vignette du fichier
feli final paper1.pdf (627.94 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03730940 , version 1 (20-07-2022)

Identifiants

  • HAL Id : hal-03730940 , version 1

Citer

Amin Feli, Rezvan Khalaji, Sakineh Kadaei, Mojtaba Banifakhr. Optimal structure for large-scale data clustering based on support vector machine and fuzzy rules. 2022. ⟨hal-03730940⟩
71 Consultations
32 Téléchargements

Partager

More