MATLAB中的径向基函数

function [net,tr]=newrb(p,t,goal,spread,mn,df)
%NEWRB Design a radial basis network.
%
%  Synopsis
%
%    net = newrb
%    [net,tr] = newrb(P,T,GOAL,SPREAD,MN,DF)
%
%  Description
%
%    Radial basis networks can be used to approximate
%    functions.  NEWRB adds neurons to the hidden
%    layer of a radial basis network until it meets
%    the specified mean squared error goal.
%
%   NET = NEWRB creates a new network with a dialog box.
%
%   NEWRB(PR,T,GOAL,SPREAD,MN,DF) takes these arguments,
%     P      - RxQ matrix of Q input vectors.
%     T      - SxQ matrix of Q target class vectors.
%     GOAL   - Mean squared error goal, default = 0.0.
%     SPREAD - Spread of radial basis functions, default = 1.0.
%     MN     - Maximum number of neurons, default is Q.
%     DF     - Number of neurons to add between displays, default = 25.
%   and returns a new radial basis network.
%
%   The larger that SPREAD is the smoother the function approximation
%   will be.  Too large a spread means a lot of neurons will be
%   required to fit a fast changing function.  Too small a spread
%   means many neurons will be required to fit a smooth function,
%   and the network may not generalize well.  Call NEWRB with
%   different spreads to find the best value for a given problem.
%
%  Examples
%
%    Here we design a radial basis network given inputs P
%    and targets T.
%
%      P = [1 2 3];
%      T = [2.0 4.1 5.9];
%      net = newrb(P,T);
%
%    Here the network is simulated for a new input.
%
%      P = 1.5;
%      Y = sim(net,P)
%
%  Algorithm<