Ability of Evolutionary and Recurrent SOM model GA-RSOM in Phonemic Recognition Optimization

International Journal of Mathematical Trends and Technology (IJMTT)          
© 2013 by IJMTT Journal
Volume-4 Issue-6                           
Year of Publication : 2013
Authors : Mohamed Salah Salhi ,Najet Arous, Noureddine Ellouze


Mohamed Salah Salhi ,Najet Arous, Noureddine Ellouze"Ability of Evolutionary and Recurrent SOM model GA-RSOM in Phonemic Recognition Optimization"International Journal of Mathematical Trends and Technology (IJMTT),V4(6):97-106. 2013. Published by Seventh Sense Research Group.

The phoneme recognition aims to process a speech signal, characterized by a non-linearity with very high dynamics, allowing to perform various tasks on an information processing machine by an operator using orally address. This paper focuses on a proposed strategy, which implements an evolutionary recurrent self organizing map(SOM) model in phonemes recognition to improve their rates. It is a hybrid model (GA-RSOM) reflecting the approaches of K-means mobile centers, the evolutionary genetic algorithm (GA) principle and the recurrent temporal appearance of Kohonen map (RSOM) to be a powerful optimization tool for phonemic recognition, even in adverse environmental conditions.


[1] Julien RACHEDI : ‘Reconnaissance et classification de phonèmes’, Laboratoire IRCAM, Paris, Aout 2005.
[2] Marie Cottrell et al : ‘Cartes auto-organisées pour l’analyse exploratoire de données et la visualisation’, Université Paris 1, 2003.
[3] Gaussier E, et al: ‘A hierarchical model for clustering and categorising documents’, Advances in Information Retrieval, Proceedings of the 24th BCS-IRSG European Colloquium on IR Research ECIR-02, Glasgow. Lecture Notes in Computer Science 2291, p.229-247, Springer, 2002.
[4] Kohonen, T.: “Self-organizing maps”, Springer, Berlin, 1995.
[5] Thomas voegtlin ‘thèse’: “Réseaux de neurones et autoréférence”, université lumière- lyon II, 2002.
[6] Mihail Crucianu : ‘ Algorithmes d'évolution pour les réseaux de neurones’, Rapport de recherche 187, Février 1997, Révisé Avril 1997, Laboratoire d'Informatique Ecole d'Ingénieurs en Informatique pour l'Industrie 64 avenue Jean Portalis 37200 Tours, France.
[7] Shimon Whiteson, Peter Stone: “Evolutionary Function Approximation for Reinforcement Learning”, to appear in Journal of Machine Learning Research, 2006. Department of Computer Sciences.University of Texas at Austin.
[8] J.H. Holland : “Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to biology, control and artificial intelligence”. MIT Press, ISBN 0-262-58111-6. 1998.
[9] Mohamed Salah Salhi, Najet Arous, Noureddine Ellouze: ‘Principal temporal extension of SOM’, journal IJSIP Koréa, laboratoire LSTS- ENIT_Tunis, septembre 2009.

—The SOM model; the recurrent SOM; the K-means mobile centers; the genetic Algorithm GA; the hybrid model (GA-RSOM).