Classification of Slovak white wines using artificial neural networks and discriminant techniques

This work demonstrates the possibility to use artificial neural networks (ANN) for the classification of white varietal wines. A multilayer perceptron technique using quick propagation and quasi-Newton propagation algorithms was the most successful. The developed methodology was applied to classify Slovak white wines of different variety, year of production and from different producers. The wine samples were analysed by the GC–MS technique taking into consideration mainly volatile species, which highly influence the wine aroma (terpenes, esters, alcohols). The analytical data were evaluated by means of the ANN and the classification results were compared with the analysis of variance (ANOVA). A good agreement amongst the applied computational methods has been observed and, in addition, further special information on the importance of the volatile compounds for the wine classification has been provided.