Radial basis function process neural network training based on generalized frechet distance and GA-SA hybrid strategy

摘要

For learning problem of Radial Basis Function Process Neural Network(RBF-PNN), an optimization training method based on GA combined with SA isproposed in this paper. Through building generalized Fr\'echet distance tomeasure similarity between time-varying function samples, the learning problemof radial basis centre functions and connection weights is converted into thetraining on corresponding discrete sequence coefficients. Network trainingobjective function is constructed according to the least square errorcriterion, and global optimization solving of network parameters is implementedin feasible solution space by use of global optimization feature of GA andprobabilistic jumping property of SA . The experiment results illustrate thatthe training algorithm improves the network training efficiency and stability.