Land evaluation based on SFAM neural network ensemble

Simplified Fuzzy ARTMAP (SFAM) neural network is characterized as self-organized feedback, incremental learning and highly complex mapping, which outperforms forward neural networks such as BP (Backward Propagation) neural network and RBF (Radial Basis Function) neural network. To overcome the influence of the ordering of training sample presentation and improve the accuracy of land evaluation, a land evaluation method using SFAM neural network ensemble was presented. Moreover, SFAM neural network, SFAM neural network ensemble, BP neural network, BP neural network ensemble, RBF neural network, and RBF neural network ensemble were used to evaluate the land in Zhongshan city of Guangdong Province, China. And the results were analyzed and compared. The experimental results demonstrate that the accuracy of land evaluation using SFAM neural network is higher than those of BP neural network and RBF neural network, respectively; for the three types of neural networks, neural network ensembles perform better than their single neural networks, respectively.