Optimized Multilayer perceptron for the geographical and genotypic classification of four genotypes of arabica coffee

Resumo

The climatic conditions of the coffee crop give special attributes to the beverage and could increase its value. However, it is essential to prove the geographical and genotype origin of the cultivar using reliable methods. Several statistical methods have been developed in an attempt to reproduce the human capability of pattern recognition. The multilayer perceptron (MLP) is an artificial neural network (ANN) with supervised learning that is widely used for pattern classification. This study aimed to develop a MLP to classify the geographic origin and the genotypic of the arabica coffee. For this purpose, spectra obtained in the Fourier transform infrared (FTIR) were analyzed using MLPs optimized by sequential simplex. The networks that used the range 1900-800 cm^-1 of the raw spectrum had lower mean squared error (MSE) and a higher percentage of correct classification for geographical (100%) and genotypic (77,74%) segmentation. After the results it was concluded that the optimized multilayer perceptrons were able to classify the samples of arabica coffee geographically. However, for genotypic classification the performance was not satisfactory. Also, the MLP developed for genotypic classification has a high number of synaptic weights, thus a large degrees of freedom database is necessary to produce a network with generalization capability. Therefore, to improve the genotype classification performance the authors suggest the use of other type of ANN and information from the near infrared.

DOI: http://dx.doi.org/10.14685/rebrapa.v3i1.96