Avaliação genética de touros da raça Tabapuã e predição da lactação de animais da raça Gir leiteiro e cruzamentos com raça Holandesa por intermédio de redes neurais artificiais

Abstract

The technique of Artificial Neural Networks (ANN) was used both for genetic evaluation of Tabapuã beef cattle body weight and for predicting total milk production of dairy cows of different genetic compositions. In the first analysis, the results from ANNs and BLUP were highly correlated although the ranked breeding values for body weight at 205 days (VG_P205_RNA) obtained by ANNs procedure and those predicted by BLUP, showed a small variation, indicating some risks in the use of ANNs for body weight genetic evaluation of Tabapuã cattle. Furthermore, insertions of new information in the data set suggested the ANNs procedure needs new training data, always dependent on BLUP. The last analysis to predict total milk production for dairy cows of different genetic composition, the milk producing ability (MPA) was calculated and compared with the observed milk production in 305 days of lactation and the results obtained by ANNs procedure. The results from ANNs procedure showed a strong association with the observed milk production and milk producing ability (MPA), the latter one being a simplified form to obtain this information