Prediction of Parent Data of Silkworm Breeding Based on Artificial Neural Network

A conventional breeding program typically deploys hundreds of crosses each year. However, only 1 or 2 percent of the combinations finally produce the desired variety. A large number of combinations are eliminated in the selection process of different breeding generations, breeding still depends largely on phenotypic selection and the experience of the breeder. The problems in traditional breeding, such as long cycles, low efficiency and poor foresight, have not been fundamentally solved. With the development of information technology and artificial intelligence, breeding with the help of modern technology has become a trend. The purpose of this study is to overcome the blindness of hybrid combinations by using computer artificial intelligence technology and establishing a simulated environment for silkworm breeding to help silkworm breeders. Based on backpropagation neural network, with the silkworm cocoon productivity as the breeding goal, the breeding model was established, the quantitative traits of more than 100 varieties were taken as samples to train the model, and the breeding simulation was carried out on this basis. By optimizing parental selection, and predicting parental selection, a hybridization test was carried out. The results showed that the difference between the predicted and tested values was 10.09% on average.