Mixed odors classification by Neural Network using Radial Basis Function

Odor research has become an attractive topic to be developed because of its potential application in advanced technologies. Thus, the classification of mixed odor becomes more important. But, classifying the mixed odors is difficult in getting high recognition. In this paper, we consider the three mixture of odor using 8 and 16 channels of sensor as main data and focusing on its classification method using neural network. The neural network used here are a Radial Basis Function Neural Network (RBFNN). K-mean clustering and Self-organizing map (SOM) are used for gaining the centers and standard deviations. We classify the three mixtures of odor into 18 and 12 classes for each data type. The result show high recognition which the accuracy rates using 16 channels of sensor are around 84.9% for 18 classes and around 93.4% for 12 classes, while the accuracy rates using 8 channels of sensor are around 88.2% for 18 classes and around 91.9% for 12 classes.