Design of Artificial Neural Networks using Evolutionary Computation

Artificial Neural Networks (ANNs) offer an attractive paradigm of computation. However, it is often hard to design good ANNs because many of the basic principles governing information processing in ANNs are difficult to understand. When complex combinations of behaviour approaches are given, and the size of the nets grows in dimension and complexity, classical solutions do not work and it is necessary to appeal to more efficient automated procedures. In this intent of automated solutions the evolutionary techniques appear, being one of their main implementations the genetic algorithms, whose search strategy is based on the population. In this paper we deal with the most important characteristics of the genetic algorithms, analysing and comparing their most important constituents, and how these algorithms can be applied to the design of ANNs.