VERIFYING WINE ORIGIN - A NEURAL-NETWORK APPROACH

Neural networks are powerful computational tools that ''learn'' with t raining examples and have the capability for extrapolating their ''kno wledge'' to new situations. Artificial back-propagation neural network s (BPNNs) were applied to two different problems of wine classificatio n. In one case, each of five 16 x 2 x 2 BPNNs was trained with a diffe rent set of 21 samples to distinguish between young red wines of two S panish Certified Brands of Origin (Ribera de Duero and Toro) on the ba sis of 15 anthocyanin contents. The networks were tested with 26 examp les not involved in the training process and gave average correct pred ictions of 88% for Toro wines and 91% for Rbera wines. In another case , five 23 x 8 x 8 BPNNs were applied to classify eight Portuguese vari etal wines of Vitis vinifera according to grape variety (Roupeiro, Man teudo, Tamarez, Rabo de Ovelha, Moreto, Trincadeira, Periquita, and Ar agonez). Percent composition of 22 free amino acids were available for 42 samples (of 7 different vintages) and were used as input data. Eac h network was trained with a different set of 26 examples and tested w ith the other 16, obtaining an average success rate of 73%. Two-layer BPNNs could be used to deduce which amino acids are characteristic of each variety of wine. In the two problems studied, the neural networks gave better predictions than linear discriminating analysis (LDA).