Title: Using genetic algorithms for automatic recurrent ANN development: an application to EEG signal classification

Authors: Daniel Rivero; Vanessa Aguiar-Pulido; Enrique Fernandez-Blanco; Marcos Gestal

Addresses: Fac. Informatica, University of A Coruña, Campus Elviña, 15071, A Coruña, Spain ' Fac. Informatica, University of A Coruña, Campus Elviña, 15071, A Coruña, Spain ' Fac. Informatica, University of A Coruña, Campus Elviña, 15071, A Coruña, Spain ' Fac. Informatica, University of A Coruña, Campus Elviña, 15071, A Coruña, Spain

Abstract: ANNs are one of the most successful learning systems. For this reason, many techniques have been published that allow the obtaining of feed-forward networks. However, few works describe techniques for developing recurrent networks. This work uses a genetic algorithm for automatic recurrent ANN development. This system has been applied to solve a well-known problem: classification of EEG signals from epileptic patients. Results show the high performance of this system, and its ability to develop simple networks, with a low number of neurons and connections.

Keywords: artificial neural networks; ANNs; genetic algorithms; GAs; signal classification; epilepsy detection; EEG signals; signal classification; electroencephalogram; epileptic patients.

DOI: 10.1504/IJDMMM.2013.053695

International Journal of Data Mining, Modelling and Management, 2013 Vol.5 No.2, pp.182 - 191

Available online: 05 May 2013 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article