Improving concentration estimation of pollutant gases by means of K-nn classification with adaptive vote

Principal Component Analysis with K-nn classification is often used in the field of electronic nose. The classification is usually based on consensual vote of the nearest neighbors. In this paper, a different decision policy is implemented, based on an adaptive vote which takes into account the structure of the-knowledge set. The method is tested on an experimental data set acquired with an array of tin oxide sensors for environmental applications.