Classification of Mixed Odors Using A Layered Neural Network

Abstract

Odor classification has been forcused due to one of five senses of human being. If we could establish the odor classification technology, we would expect various new technology since human being requires five sences to acheive higher quality information processing and sophistcated decision making. For example, we could expect the odor classification and odor synthesis, which enable us to achieve odor communication technology. Furthermore, the odor classification would be applicable to keep the society safe by detecting the dangerous odors and to make our life more satisfactory by using additional odor information. In this paper we develop an electronic nose using a neural network. The neural network is a multi-layered neural network based on the gradient method. After classifying the various odors, we consider the classification in case that mixed odors are measured. To improve the classification accuracy we adopt a genetic algorithm to find a reduction factor to separate two mixed odors.