Application of artificial intelligence methods for bovine gender prediction

Abstract

This study investigates determining the gender of calves using some artificial intelligence (AI) techniques. Gender identification is important in animal breeding, focusing on the desired outcome and planning. The data used to determine the gender of calves were the speed, magnitude, and density of the bull's semen. The analysis of the related studies showed that there was not a study on gender prediction of bovine with the application of AI methods. In this study, fuzzy logic (FL), artificial neural networks (ANN), support vector machines (SVM), and random forests (RF) were used. The efficiency of these approaches was verified by statistical analysis parameters such as accuracy, specificity, sensitivity (recall), precision, and F-score. The FL, ANN, SVM, and RF models had 84%, 96%, 97%, 99% accuracy, 93.75%, 96.88%, 100%, 100% sensitivity, 66.66%, 94.44%, 92.31%, 97.30% specificity, 83.33%, 96.88%, 95.31%, 98.44% precision results, respectively. Application of these AI techniques for prediction bovine gender proves that these methods may be used by semen breeders as supporting information tools. In particular, it was observed that the RF method yielded the highest accuracy results.