Neural network based electronic nose for discrimination of fish sauces

Fish sauce is one of the signature condiments in various cuisines in many countries. In this study, fish sauces are successfully discriminated depending on their quality indicated by the level of Total Nitrogen (TN) content. We introduce an electronic nose (E-nose) technology to measure the odor of fish sauce. Feature extraction methods are also performed in order to obtain information relevant to the odor discrimination task. Four features based on both steady-state and transient responses are extracted from each response signal obtained from the E-nose. The principal component analysis (PCA) is also introduced to reduce the dimensionality of feature vector for avoiding the curse of dimensionality. The backpropagation (BP) learning algorithm is then used for the discrimination. Using this method, we achieve high discrimination performance to the fish sauces.