Selective approach for neural network ensemble based on network clustering technology and its application

To solve the difficult issue of crop precise fertilization, a neural network ensemble approach is proposed, which is based on the complex network clustering algorithm. In this approach, first, the method of sampling with replacement is adopted to produce the number of neural network units. Then, Yang's network clustering Forward Error Correction (FEC) algorithm is used to select the networks with high precision and great diversity. Third, the selected networks are ensembled separately with linear weighted ensemble method and nonlinear ensemble method. Finally, the prediction result is achieved by amalgamating the results produced by the two ensemble methods. Test results, carried out at the No.7 corn plot in Yushu city in 2008, reveal that it proposed approach is better than the traditional fertilization models, linear weighted ensemble model and nonlinear ensemble model. In addition, the ability of generalization of the approach is strong.