Odor classification using Support Vector Machine

Abstract

This paper discusses about the process of classifying odor using Support Vector Machine. The training data was taken using a robot that ran in indoor room. The odor was sensed by 3 gas sensors, namely: TGS 2600, TGS 2602, and TGS 2620. The experimental environment was controlled and conditioned. The temperature was kept between 27.5 0C to 30.5 0C and humidity was in the range of 65%–75 %. After simulation testing in Matlab, the classification was then done in real experiment using one versus others technique. The result shows that the classification can be achieved using simulation and real experiment.