Performance Analysis of Neural Networks Training using Real Coded Genetic Algorithm

Abstract

Multilayer perceptrons (MLPs) are widely used for pattern classification and regression problems. Backpropagation (BP) algorithm is known technique in the training of multilayer perceptrons. However for its optimum training convergence, the learning and momentum parameters need to be tuned on trial and error method. Further, sometimes the backpropagation algorithm fails to achieve global convergence. To alleviate these problems we suggest a genetic algorithm based training for MLP network. Both binary coded and real coded genetic algorithm are used and a comparative training performance analysis has been studied. It is observed from simulation results that both the schemes outperform backpropagation algorithm and achieve global convergence. Further the real coded GA based training shows a faster convergence than binary coded GA based training. For simulation datasets are taken from UCI based machine learning repository. Hence real-coded genetic algorithm finds an alternative for back propagation based training algorithm.

A Hybrid Differential Evolution and Back-Propagation Algorithm for Feedforward Neural Network Training

“…In the past few years, pattern classification using feed-forward neural networks (FNN), in specific, multilayer perceptron (MLP) is considered as promising neural network model [2] due to its capability to classify the real world complex problem without prior knowledge on the problem domain. In classification, first the network is trained on a set of paired data from the dataset to evolve a set of free network parameters that is, selection of optimal weights of the network weights and second, then the network is ready to test a new set of data [5].Training the weights of the multilayer perceptron can be considered as an optimization problem in which the network weights are optimized.…”

“…J. D. Schaffer et al [17] wrote a survey on combinations of genetic algorithms and neural networks. The performance of back-propagation algorithm compared with binary and real-coded genetic algorithms for training multi-layer perceptron has been referred in [2,3]. Iloren et al [1] compared back-propagation with differential evolution for neural networks training.…”