An Optimized Neural Network Classification Method Based on Kernel Holistic Learning and Division

An optimized neural network classification method based on kernel holistic learning and division (KHLD) is presented. The proposed the learned radial basis function (RBF) as research object. here can be considered a subspace region consisting of same pattern category in training sample space. By extending space original instances, relevant information between instances obtained from subspace, classifier’s boundary far instances; thus, robustness generalization performance classifier are enhanced. In concrete implementation, new vector generated within each RBF according to instance optimization screening characterize KHLD. Experiments artificial datasets several UCI benchmark show effectiveness our method.