ИСПОЛЬЗОВАНИЕ НЕЙРОННЫХ СЕТЕЙ В ГЕНОТИПИРОВАНИИ

Abstract

For effective breeding work, methods like BLUP (BLUP is the best linear unbiased predictor) and their comparison with breeding indices are increasingly being introduced. At the same time, the accuracy of information about the kinship relationships of animals increases, but at the same time the mathematical processing of such data becomes more complicated, requiring deeper analysis and comparison of large data sets, including using machine learning. Particular importance is given to the search for deep machine learning opportunities with the inclusion of information about the genotypes of the maximum number of animals of the same and different genealogical branches of the pedigree in the breeding assessment, which is in demand by both academic circles and breeding farms. This article describes and analyzes the relevance of neural networks in identifying the genetic potential of populations. In addition, proposed approaches to regulate the effect of breeding on various breeding characteristics included in the functions of the neural network.