Smells classification for human breath using a layered neural network

Abstract

Progress of sensor technology enables us to measure smells although it is based on chemical reactions. We have developed the smell classification for various subjects using layered neural networks by training a special smell. But we must learn many smells by repeating the same process and it is endless jobs since too many smells exist in the world. In case of a breath smell, several molecules are mixed. Therefore, if we can train basic components of the breath and mixtures are estimated by combining the basic components which consist the breath, it is preferable. In this paper, we develop mixed smell classification after training a neural network for each component by using a genetic algorithm to find a reduction factor from the measurement data which show the maximum value of the output of a layered neural network.