Several gradient-based approaches such as back-propagation, conjugate gradient and Levenberg Marquardt (LM) methods have been developed for training the neural network (NN) based systems. Still, in some situations, like multimodal cost function, these procedures may lead to some local minima, therefore, the evolutionary algorithms (EAs) based procedures were considered as a promising alternative. In this paper we focus on a memetic algorithm based approach for training the neural networks. We use the memetic algorithm (MA) for training the multilayer perceptrons as applied to nonlinear system identification. The proposed memetic algorithm is an alternative to gradient search methods, such as back-propagation, which have inherent limitations of many local optima. Here we have studied the identification of a nonlinear system using five different algorithms namely back-propagation (BP), genetic algorithm (GA), differential evolution (DE), genetic algorithm back-propagation (GABP), along with the proposed differential evolution plus back-propagation (DEBP) approaches. In the proposed system identification scheme, we have exploited two global search methods namely genetic algorithm (GA), and differential evolution (DE). These two global search methods have been hybridized with the gradient descent method i.e. the back propagation (BP) algorithm. The local search BP algorithm is used as an operator for genetic algorithm and differential evolution. These algorithms have been tested on a standard benchmark problem for nonlinear system identification to prove their efficacy.
[1]
C. S. George Lee,et al.
Neural fuzzy systems: a neuro-fuzzy synergism to intelligent systems
,
1996
.
[2]
Terence C. Fogarty,et al.
A Genetic Algorithm with Variable Range of Local Search for Tracking Changing Environments
,
1996,
PPSN.
[3]
Rainer Storn,et al.
Differential Evolution – A Simple and Efficient Heuristic for global Optimization over Continuous Spaces
,
1997,
J. Glob. Optim..
[4]
Peter Merz,et al.
Memetic algorithms for combinatorial optimization problems : fitness landscapes and effective search strategies
,
2006
.
[5]
Michael N. Vrahatis,et al.
Memetic particle swarm optimization
,
2007,
Ann. Oper. Res..
[6]
Madan M. Gupta,et al.
Static and Dynamic Neural Networks: From Fundamentals to Advanced Theory
,
2003
.
[7]
Xin Yao,et al.
Evolutionary Artificial Neural Networks
,
1993,
Int. J. Neural Syst..
[8]
Richard D. Braatz,et al.
On the "Identification and control of dynamical systems using neural networks"
,
1997,
IEEE Trans. Neural Networks.