Classification of soybean cultivars by means of artificial neural networks

The cultivation of soy has an economic importance for the Brazilian agricultural scenario. aim this study was to establish a network architecture classification soybean genotypes, by means morphological characters measured in juvenile phase plant, and finally compare results obtained through Artificial Neural Network (ANN) Anderson Discriminant Analysis. analyzed plants 10 conventional cultivars initial stages development (V1, V2 V3 stages). experiment carried out randomized block design with 5 replications, experimental unit represented 9 plants. data were submitted Analysis multilayer Perceptron ANN, 1 or 2 hidden layers. To analyze homogeneity variance covariance matrix, Box’s M-Test adopted Program R, at 5% significance level. An input layer, one two layers, output layer formed ANN architecture. 5-fold cross validation used verify efficiency discriminant functions also analysis. Subsequently, apparent error rate (AER) obtained. indicated inhomogeneity matrices, which need perform Anderson's Quadratic ANNs presented lower when compared artificial neural sufficient cultivars.