Machine learning | matlab implementation of RBF radial basis function neural network Newrbe parameter setting

machine learning | MATLAB Realization RBF Radial basis function neural network newrbe Parameter setting

Basic introduction

RBF Shenjiang network is a three-layer neural network , It includes the input layer 、 Cryptic layer 、 Output layer . The transformation from input space to hidden space is nonlinear , The transformation from hidden space to output space is linear .
RBF The neural network adopts multi-dimensional space interpolation fitting technology , And BP The difference between error feedback neural networks is ：
1、 There is no error feedback process and the weight is calculated by the least square ;
2、 The input layer is the distance between each sample and the cluster center , Cluster centers are selected from samples ;
3、 The maximum number of nodes in the hidden layer is the maximum dimension of the sample data , Hidden nodes are listed in order from 0 Add cluster center calculation w And error MSE, The order of adding cluster centers is determined by K-means Basic function value determination .

Function description

• The network structures,

%--