TUNING THE PARAMETERS OF AN ARTIFICIAL NEURAL NETWORK USING CENTRAL COMPOSITE DESIGN AND GENETIC ALGORITHM

An artificial Neural Network (ANN) is an efficient approach applied to solving a variety of problems. The main problem in using ANN is parameter tuning, because there is no definite and explicit method to select optimal parameters for the ANN parameters. In this study, three artificial neural network performance measuring criteria and also three important factors which affect the selected criteria have been studied. Moreover, central composite design has been used to design experiments and also analyze network behavior according to identified parameters, by using the overall desirability function. Then the Genetic Algorithm has been proposed to find optimal parameter status. For this purpose, the proposed method has been illustrated by the numerical example of a well known mathematical function. The results show that the designed ANN, according to the proposed procedure, has a better performance than other networks by random selected parameters and also parameters which are selected by the Taguchi method. In general, the proposed approach can be used for tuning neural network parameters in solving other problems.

[1]  Douglas C. Montgomery,et al.  Modified Desirability Functions for Multiple Response Optimization , 1996 .

[2]  James Tannock,et al.  The optimisation of neural network parameters using Taguchi’s design of experiments approach: an application in manufacturing process modelling , 2005, Neural Computing & Applications.

[3]  Anders Krogh,et al.  Introduction to the theory of neural computation , 1994, The advanced book program.

[4]  C George,et al.  A Balancing Act: Optimizing a Product's Properties , 1994 .

[5]  Bong-Jin Yum,et al.  Robust design of multilayer feedforward neural networks: an experimental approach , 2004, Eng. Appl. Artif. Intell..

[6]  Mohammad Bagher Menhaj,et al.  Training feedforward networks with the Marquardt algorithm , 1994, IEEE Trans. Neural Networks.

[7]  Daniel C. St. Clair,et al.  Using Taguchi's method of experimental design to control errors in layered perceptrons , 1995, IEEE Trans. Neural Networks.

[8]  Steve Y. Chiu,et al.  Fine-tuning a tabu search algorithm with statistical tests , 1998 .

[9]  John R. Koza,et al.  Genetic programming - on the programming of computers by means of natural selection , 1993, Complex adaptive systems.

[10]  Shih-Ming Yang,et al.  Neural Network Design by Using Taguchi Method , 1999 .

[11]  In-Jun Jeong,et al.  An interactive desirability function method to multiresponse optimization , 2009, Eur. J. Oper. Res..

[12]  Melanie Mitchell,et al.  An introduction to genetic algorithms , 1996 .

[13]  B. S. Lim,et al.  Optimal design of neural networks using the Taguchi method , 1995, Neurocomputing.

[14]  Martin T. Hagan,et al.  Neural network design , 1995 .

[15]  Chang Wook Ahn,et al.  Advances in Evolutionary Algorithms: Theory, Design and Practice , 2006, Studies in Computational Intelligence.

[16]  Jonas Sjöberg,et al.  Efficient training of neural nets for nonlinear adaptive filtering using a recursive Levenberg-Marquardt algorithm , 2000, IEEE Trans. Signal Process..

[17]  Michael Sylvester Packianather,et al.  Optimizing the parameters of multilayered feedforward neural networks through Taguchi design of experiments , 2000 .

[18]  Russell C. Eberhart,et al.  Parameter Selection in Particle Swarm Optimization , 1998, Evolutionary Programming.

[19]  R. H. Myers,et al.  Response Surface Alternatives to the Taguchi Robust Parameter Design Approach , 1992 .

[20]  Paul J. Werbos,et al.  Backpropagation Through Time: What It Does and How to Do It , 1990, Proc. IEEE.

[21]  G. Derringer,et al.  Simultaneous Optimization of Several Response Variables , 1980 .

[22]  Nadir Yayla,et al.  The investigation of model selection criteria in artificial neural networks by the Taguchi method , 2007 .

[23]  Xun Sun,et al.  Optimizing the novel formulation of liposome-polycation-dna complexes (lpd) by central composite design , 2004, Archives of pharmacal research.