Chapter 6 - Evolutionary Algorithms and Neural Networks

Abstract

This chapter deals with a few applications of evolutionary algorithms (EAs) and combinations of EAs and neural networks (NNs). Artificial neural networks (ANNs) serve as strong predictive models. An ANN can uncover complex relationships; however, it provides little insight into the underlying mechanisms that describe the relationship. Evolutionary algorithms are not only robust methods of exploring complex solution spaces, but also provide some insights into revealing the mechanisms relating data items. While NNs are metaphorically based on learning processes in individual organisms, EAs are inspired by evolutionary changes in populations of individuals. Relative to neural nets, evolutionary algorithms have only recently gained wide acceptance in academic and industrial circles. Evolutionary algorithms describe computer-based problem solving techniques that use computational models of some of the known mechanisms of evolution as key elements in their design and implementation. The performance of ANNs is critically dependent on factors such as the choice of primitives (neurons), network architectures, and the learning algorithm used.