Using evolutionary algorithms for the structural optimization of an artificial neural network performing the analysis of electronic nose data

In this contribution we present how artificial neural networks (ANNs) can be optimized by using genetic algorithms (GAs) and evolution strategies (ESs). Our exemplary application is an ANN designed to analyse the data obtained by a gas sensor array, a so-called electronic nose, which is a gas detection system using low-level sensors together with high-level data evaluation. The evolutionary optimized ANN gives a better performance of the electronic nose data with smaller networks than our network so far.