فهرست مقالات generation interval

• دسترسی آزاد مقاله
  • صفحه چکیده
  • متن کامل
  
  1 - Estimating Heritabilities and Breeding Values for Real and Predicted Milk Production in Holstein Dairy Cows with Artificial Neural Network and Multiple Linear Regression Models
  M. Nosrati S.H. Hafezian M. Gholizadeh
  The success of a dairy herd depends on milk production. Prediction of future records can reduce recording time, accelerate the computation of genetic evaluations, decrease generation interval, and increase genetic progress. Multiple linear regression (MLR) is the most c چکیده کامل

  The success of a dairy herd depends on milk production. Prediction of future records can reduce recording time, accelerate the computation of genetic evaluations, decrease generation interval, and increase genetic progress. Multiple linear regression (MLR) is the most common prediction method. However, artificial neural networks (ANN) can handle complex linear and non-linear functions to solve a wide range of prediction problems. In this study, MLR and ANN models were applied to the prediction of 305-day milk production in the first and second lactations of dairy cows using variables related to milk production, test-day records and estimated breeding values (EBVs). The 305-day first lactation records were also used to predict 305-day second lactation records. ANN and MLR predictions were compared in terms of accuracy and efficiency. Dairy records from 7856 dairy cows in two herds were used in this research. The best ANN model was a multilayer perceptron with a back-propagation learning algorithm. Results showed that ANN and MLR predicted values were acceptable. However, ANN prediction accuracies for 305-day milk production in the first and second lactations were higher than those of MLR. Correlation coefficients between real and predicted 305-day milk production records in the first and second lactations ranged from 0.88 to 0.96 for ANN and from 0.66 to 0.89 for MLR. Adding test-day records and EBVs for 305-day milk production in the first lactation to the set of independent variables used to predict 305-day milk production in the second lactation increased more the prediction efficiency of ANN than MLR. Thus, ANN could be used to decrease the interval between collecting records and computing animal breeding values. In addition, real data and ANN-predicted data from the first lactation were used to compute EBVs. The correlation between EBVs with real and predicted data was 0.93. Results suggested that ANN could be useful for predicting complex traits using high dimensional genomic information. پرونده مقاله