Program of Masters & PhD in Engineering Faculty of Engineering •••••••••• Paola.tellez911@gmail.com National Autonomous University of Mexico Mexico
Genetic algorithms are an alternative for optimizing the performance of recurrent neural networks by searching for the best number of hidden neurons. Genetic algorithms have been used for heuristics, but never to find the best number of hidden neurons in a recurrent neural network. Three architectures of recurrent neural networks were used to measure the performance with spoken Spanish digits. 13 was the number of hidden neurons used for a Jordan network to give the best performance. This number lets to optimize resources in hardware implantation.
TITLE - Number Of Hidden Neurons In A Recurrent Neural Network Estimation AUTHOR - Paola Tellez ••••• IJSER Edition - January 2013 UNIVERSITY - National Autonomous University of Mexico GUIDE NAME - Jose de Jesus Acosta Flores Angel Kuri Morales Manuel Ordorica Melllado
