-
L. P. O. Sousa
Departamento de Automática, Universidade Federal de Lavras, Minas Gerias
-
D. D. Ferreira
Departamento de Automática, Universidade Federal de Lavras, Minas Gerias
-
D. S. Correa
Empresa Brasileira de Pesquisa Agropecúaria, EMBRAPA, São Paulo
-
J. E. Oliveira
Departamento de Engenharia de Materiais, Universidade Federal de Lavras, Minas Gerias
Keywords:
E-tongue, classification model, artificial neural networks, principal curves
Abstract
Devices that resemble biological recognition systems, that is, inspired by the function- ing of human organs, have made important advances in recent decades. Therefore, interest in the concept of electronic tongues has increased considerably in recent years. This study uses this type of device in order to identify different samples of flavor enhancers (Glutamate, Guanylate and Inosinate). The applied methodology employs different computational techniques for the identification such as Principal Curves and three types of Artificial Neural Networks. The best results reached 95,4% accuracy through models implemented with Self Organization Maps.
