Olfactory classification via interpoint distance analysis

Detection of the presence of a single prespecified chemical analyte at low concentration in complex backgrounds is a difficult application for chemical sensors. The article considers a database of artificial nose observations designed specifically to allow for the investigation of chemical sensor data analysis performance on the problem of trichloroethylene (TCE) detection. We consider an approach to this application which uses an ensemble of subsample classifiers based on interpoint distances. Experimental results are presented indicating that our nonparametric methodology is a useful tool in olfactory classification.