Independent Component Regression for the Development of Prediction Model for Analysis of Electronic Tongue Response

The paper introduces the application of Independent Component Regression (ICR) for the development of a model to predict the biochemical information from a liquid sample using an electronic tongue. We have taken the case of tea samples as analyte and the prediction model is used for the estimation of total theaflavin (TF) content in tea samples. The ICR is done using the data obtained from an electronic tongue in which different tea samples are perturbed by a voltage signal and the corresponding current response is recorded. A feature set is obtained from these responses which contain important biochemical taste determining information of tea. The estimated total TF content values of different tea are compared with that of the actual TF value obtained from the conventional laboratory method. Also, the ICR results are compared with that of a standard data analysis method called Principal Component Regression (PCR).