论文部分内容阅读
本文提出了一种优化径向基函数神经网络 (RBFNN)结构和参数的方法 ,该方法包括两个过程 :训练和进化 .训练采用梯度下降法学习 RBFNN的中心 ,宽度和输出权值 ;进化采用二进制编码的遗传算法 (GA )学习 RBFNN的结构 ,适应度函数是基于信息论中最小描述长度 (MDL)原理的目标函数 .函数逼近仿真实验证明了该方法比其他方法鲁棒性强 ,所得到的网络结构简单 .
This paper presents a method to optimize the structure and parameters of Radial Basis Function Neural Network (RBFNN). The method includes two processes: training and evolution, and the training adopts the gradient descent method to study the center, width and output weights of RBFNN. Binary-coded genetic algorithm (GA) learns the structure of RBFNN, and fitness function is the objective function based on the principle of least description length (MDL) in information theory. The approximation of the simulation results shows that the proposed method is more robust than other methods, Network structure is simple.