The International Arab Journal of Information Technology (IAJIT)


Training of Fuzzy Neural Networks via Quantum- Behaved Particle Swarm Optimization and Rival

Saeed Farzi,
There  are  some  difficulties  encountered  in  the  appl ication  of  fuzzy  Radial  Basis  Function  (RBF)  neural   network.  One of them is how to determine the number of hidde n rule neurons and another difficulty is about interpretability. In order to  overcome these difficulties, we have proposed a fuz zy neural network based on RBF network and takagi-s ugeno fuzzy system.  We  have  used  a  new  structure  of  fuzzy  RBF  neural  ne twork,  which  has  been  proved  that  it  is  better  than  other  structures  in  term  of  interpretability.  Our  model  also  use  a  Riva l  Penalized  Competitive  Learning  (RPCL)  and  a  swarm   based  algorithm  called  Quantum-behaved  Particle  Swarm  Optimization  (QPSO)  to  determine  design  parameters  of  hidden  layer  and  design  parameters  of  output  layer,  respectively.  RPCL  is  t he  best  clustering  algorithm  that  is  introduced  so far.  The  Particle  Swarm  Optimization  (PSO)  is  a  well-known  population-based   swarm  intelligence  algorithm.  The  QPSO  is  also  pro posed  by  combining  the  classical  CPSO  philosophy  and  quantum   mechanics  to  improve  performance  of  PSO.  We  have  c ompared  the  performance  of  the  proposed  method  with  gradient  ba sed  method.  Simulation  results  of  nonlinear  function  approximation  demonstrate the superiority of the proposed method  over gradient based method. 

[1] Bao H., Huang H., and Li X., A Study of the T# S Fuzzy RBF Neural Network, Journal of Huazhong University of Science and Technology , vol. 27, no. 1, pp. 11#13, 1999.

[2] Jang S. and Sun T., Functional Equivalence between Radial Bases Functions and Fuzzy Inference Systems, IEEE Transactions on Neural Networks , vol. 4, no. 1, pp. 156#158, 1993.

[3] Jang S. and Sun T., Neuro#Fuzzy Modeling and Control, in Proceedings of the IEEE , pp. 378# 406, 1995.

[4] Jin Y. and Sendhoff B., Extracting Interpretable Fuzzy Rules from RBF Networks, Neural Processing Letters , vol. 17, no. 2, pp. 149#164, 2003.

[5] Jin Y., Fuzzy Modeling of High#Dimensional Systems Complexity Reduction and Interpretability Improvement, IEEE Transactions on Fuzzy Systems , vol. 8, no. 2, pp. 212#221, 1995.

[6] Kennedy J. and Eberhart C., Particle Swarm Optimization, in Proceedings of IEEE International Conference on Neural Network , Australia, pp. 1942#1948, 1995.

[7] Linkens A. and Chen Y., Input Selection and Partition Validation for Fuzzy Modeling Using Neural Network, Journal of Fuzzy Sets Systems , vol. 107, no. 3, pp. 299#308, 1999.

[8] Li W. and Hori Y., An Algorithm for Extracting Fuzzy Rules Based on RBF Neural Network, IEEE Transactions on Industrial Electronics , vol. 53, no. 4, pp. 1269#1276, 2006.

[9] Rosenblatt R., Principles of Neurodynamics , Spartan Books, New York, 1959.

[10] Sun J., Feng B., and Xu B., Particle Swarm Optimization with Particles Having Quantum Behavior, in Proceedings of Congress on Evolutionary Computation , USA, pp. 325#331, 2004.

[11] Sun J. and Xu B., A Global Search Strategy of Quantum#Behaved Particle Swarm Optimization, in Proceedings of IEEE Conference on Cybernetics and Intelligent Systems , Singapore, pp. 111#116, 2004.

[12] Sun J. and Xu B., Adaptive Parameter Control for Quantum#Behaved Particle Swarm Optimization on Individual Level, in Proceedings of IEEE International Conference on Systems , USA, pp. 3049#3054, 2005.

[13] Sun J. and Xu W., Parameter Selection of Quantum#Behaved Particle Swarm Optimization, in Proceedings of Advances in Natural Computation , Heidelberg, pp. 543#552, 2005.

[14] Takagi T. and Sugeno M., Fuzzy Identification of Systems and Its Applications to Modeling and Control, in Proceedings of IEEE Transactions on Systems , pp. 116#132, 1985.

[15] Wang X. and Mendel J., Generating Fuzzy Rules by Learning from Examples, IEEE Transactions on Systems , vol. 22, no. 6, pp. 1414#1427, 1992.

[16] Xu L. and Krzyzak A., Rival Penalized Competitive Learning for Clustering Analysis, RBF Net, and Curve Detection, IEEE Transactions on Neural Networks , vol. 4, no. 4, pp. 636#649, 1993.

[17] Zadeh A., Outline of A New Approach to the Analysis of Complex Systems and Decision Processes, IEEE Transactions on Systems , vol. 3, no. 1, pp. 28#44, 1973. Saeed Farzi received his BS in computer engineering from Razi University, Iran in 2004, and his MS in artificial intelligence from Isfehan University, Iran in 2006. He is a faculty member at the Department of Computer Engineering, Islamic Azad University#beranch of Kermanshah, Iran. His current research interests include artificial intel ligence, neural network, soft computing and grid computing.