Data from 19,240 Tabapuã animals from 152 farms located in different states of Brazil, born from 1976 to 1995, were used to predict the genetic value of body weight at 205 days of age (BV_P205) of Tabapuã beef cattle using Artificial Neural Networks (ANN) and LM algorithm - Levenberg Marquardt training for data entry. Due to the use of networks with supervised learning, the predicted breeding values for P205 from BLUP were used as desired output. The breeding values for P205 obtained from RNA and those predicted by BLUP were highly correlated. The ranked breeding values for body weight at 205 days through RNA and those predicted by BLUP (VG_P205_RNA) showed a variation in the classification of animals indicating risks in the use of ANNs procedure for genetic evaluation of this trait. Insertions of new animals require new training data always dependent on BLUP.
beef cattle; artificial neural networks; genetic evaluation; best linear unbiased predictor
