Development of Breeding Management Support System Using Neural Network

Research Project

Project/Area Number	07806033
Research Category	Grant-in-Aid for Scientific Research (C)
Allocation Type	Single-year Grants
Section	一般
Research Field	生物環境
Research Institution	University of Tsukuba
Principal Investigator	SATAKE Takaaki Univ.of Tsukuba, Inst.of Agr.and Forest Eng., Associate Professor, 農林工学, 助教授 (00170712)
Project Period (FY)	1995 – 1997
Project Status	Completed (Fiscal Year 1997)
Budget Amount *help	¥400,000 (Direct Cost: ¥400,000) Fiscal Year 1997: ¥200,000 (Direct Cost: ¥200,000) Fiscal Year 1996: ¥200,000 (Direct Cost: ¥200,000)
Keywords	Breeding Management / Bred Swine / Growth Prediction / Finish Daily Age / Neural Network Model / Linear Multiple Regression Model / ニューラルネットワークモデル / 生長予測
Research Abstract	In order to obtain basic information for the construction of a livestock management support system using neural network, a simulation of growth prediction of bred swine was conducted using 3 layred back-propagation network. In addition to this, a simulation of growth prediction using linear multiple regression model was also carried out to compare with the result of the simulation using neural network model. As a first step of the study, seven growth prediction factors such as genealogy, sex, birth weight, intake of digestive energy, average house temperature and others were selected, and the simulation of growth prediction using neural network was carried out by changing the number of hidden units and learning cycles of the network. As the results of these simulations, the percentage of apparent prediction level of neural network model was about 1-11% higher than that of linear multiple regression model, and it was clarified that the use of neural network was effective for growth prediction of swine.