عنایت رحمت نژاد

Dietary supplements such as herbal additives affect performance of broiler chicken. This study aims to present the empirical techniques for predicting live weight (LW) of broiler chickens when fed to six dietary treatments including (1) basal diet, (2) organic acids, (3) shallot, (4) yarrow, (5) mixture of shallot and yarrow, and (6) antibiotic from 0 to 42 d of age. The information about feeding these six diets containing different supplements between 0 and 42 days of life were used to predict LW of broilers. A neural network with different structures was developed by using MATLAB software, and genetic algorithm (GA) was employed to determine the optimal value for the initial weights of neural networks. Dietary treatments and initial weight (at 1 d) were used as input variables and LW of treatments was output variable. The best model of artificial neural network- genetic algorithm (ANN-GA) was determined based on root mean square error (RMSE). The best selected ANN-GA has shown desirable results (RMSE values 66.8 grams and R 2 coefficient 0.94569). Based on the results of this study, ANN-GA is an appropriate, cheap, and reliable tool to predict LW of broiler.