function xNewff
% xNewff.m
% 本程序用newff()創(chuàng)建兩層前向網(wǎng)絡(luò)，并用sim()對網(wǎng)絡(luò)進(jìn)行仿真
%
%   Author: HUANG Huajiang
%   Copyright 2002 UNILAB Research Center,
%   East China University of Science and Technology, Shanghai, PRC
%   $Revision: 1.0 $  $Date: 2003/01/10 $
%
%   [Ref] MATLAB demo, Mathworks Co.

clear all
clc

% 輸入樣本p和目標(biāo)t
p = -1:.1:1;
t = [-.9602 -.5770 -.0729  .3771  .6405  .6600  .4609 ...
          .1336 -.2013 -.4344 -.5000 -.3930 -.1647  .0988 ...
          .3072  .3960  .3449  .1816 -.0312 -.2189 -.3201];

% 用newff()創(chuàng)建兩層前向網(wǎng)絡(luò)，網(wǎng)絡(luò)輸入范圍是[-1 1]，第一層有10個(gè)
% TANSIG神經(jīng)元，第二層有1個(gè)PURELIN神經(jīng)元。訓(xùn)練函數(shù)為TRAINLM()
net = newff([-1 1],[10 1],{'tansig' 'purelin'},'trainlm');

% 對前向網(wǎng)絡(luò)進(jìn)行仿真，并將網(wǎng)絡(luò)輸出值對目標(biāo)值作圖
y = sim(net,p)
plot(p,t,'o',p,y,'x')  %
title('訓(xùn)練向量 (訓(xùn)練1個(gè)epoch)');
xlabel('輸入向量 p');
ylabel('目標(biāo)向量 t');

%   Here the network is trained for 50 epochs.
%   Again the network's output is plotted.
net.trainParam.epochs = 50;
net = train(net,p,t);
y = sim(net,p)
plot(p,t,'o',p,y,'*-')
title('訓(xùn)練向量 (訓(xùn)練50個(gè)epoch)');
xlabel('輸入向量 p');
ylabel('目標(biāo)向量 t');