The optimization of radial basis probabilistic neural networks based on genetic algorithms
Lin Guo, Do-Shuang Huang, Wenbo Zhao · 2004
In this paper, a genetic algorithm (GA) is introduced into optimizing the radial basis probabilistic neural networks (RBPNN). The encoding method proposed in this paper involves not only the number and the locations of selected hidden centers but also the shape parameter of the Gaussian kernel function. We use the telling-two-spirals-apart problem as an example to validate the genetic algorithm for optimizing the RBPNN. Consequently, we obtain an optimal interval of the shape parameter of the kernel function for this problem except the reduced RBPNN structure (including the optimal number of the hidden centers and their optimal locations). The experimental results show that with the shape parameters in the optimal interval and with the optimized hidden centers the designed network is not only parsimonious but also of better generalization performance.