The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON INFORMATION SCIENCE AND APPLICATIONS
Transactions ID Number: 88-113
Full Name: Iulian-Constantin Vizitiu
Position: Associate Professor
Age: ON
Sex: Male
Address: George Cosbuc Avenue 81-83
Country: ROMANIA
Tel:
Tel prefix:
Fax:
E-mail address: iulian_vizitiu@yahoo.com
Other E-mails: vic@mta.ro
Title of the Paper: Training of RFB neural networks using a full-genetic approach
Authors as they appear in the Paper: Iulian-Constantin VIZITIU, Petricã CIOTÎRNAE, Teofil OROIAN, Adrian RADU,Florin POPESCU, Cristian AVRAM
Email addresses of all the authors: vic@mta.ro,ciotirnae@mta.ro,oroiant@mta.ro,i_radu@mta.ro,fdpopy@mta.ro,cavram@mta.ro
Number of paper pages: 10
Abstract: The efficiency of pattern recognition (PR) systems using RBF neural networks to implement their recognition function, depends a lot by the training algorithms of these neural networks and especially, by the specific techniques (e.g., supervised, clustering techniques etc.) used for RBF center positioning. Having as starting point the basic property of genetic algorithms (GA) to represent global searching tools, a full-genetic approach to assure optimization both connectivity and neural weights of RBF networks is proposed. In order to confirm the broached theoretical aspects and based on real pattern recognition task, a comparative study (as performance level) with others standard RBF training methods and SART neural network is also indicated.
Keywords: RBF neural networks, clustering techniques, genetic algorithms
EXTENSION of the file: .pdf
Special (Invited) Session: A genetic procedure used to train RFB neural networks
Organizer of the Session: 632-298
How Did you learn about congress: cmolder@mta.ro
IP ADDRESS: 85.120.255.210
