chapter nine – Applications

Publisher Summary This chapter presents the practical world of engineering and shows some applications of particle swarm optimization. Evolutionary computation methodologies have generally been applied to three main attributes of neural networks: network connection weights, network architecture, and network learning algorithms. The chapter summarizes some of the advantages and disadvantages that have been discussed in the literature and experienced with respect to the use of evolutionary computation techniques with artificial neural networks. It highlights the successes and examines issues that should be addressed in order to make further progress. The performance of each particle is measured according to a predefined fitness function, which is related to the problem to be solved. The chapter presents a new methodology using particle swarm optimization for evolving neural network weights and simultaneously indirectly evolving network architecture. . This chapter also describes the application of particle swarm optimization to evolve a neural network that classifies human tremor (Parkinson's disease or essential tremor) versus normal subjects. . The relatively small size of the data set indicates the need for further testing and development.