clear
% 原始数据导入
iris = load('C:\MATLAB7\work\模式识别\iris.txt');
M=30;
for k=1:3
for i = 1:M
for j = 1:4
x1(i+(k-1)*M,j) = iris(i+(k-1)*50,j+1);
end
d1(i+(k-1)*M,1)=abs(mod(k,2)-0.1);
d1(i+(k-1)*M,2)=abs(floor(k/2)-0.1);
end
end
x2(:,1:4)=x1;
x2(:,5:6)=d1;
%训练样本顺序打乱
xd=x2(randperm(numel(x2)/6),:);
x=xd(:,1:4);
d=xd(:,5:6);
[P,minp,maxp] = premnmx(x);%归一化
P=P';
T=d';
net=newff(minmax(P),[12,2],{'logsig' 'logsig'},'trainrp');
%purelin纯线性函数 trainrp用于模式识别速度较快
net.trainparam.epochs=2000;
net.trainparam.goal=0.0045;
net=train(net,P,T);
sim(net,P)
N=50-M;
for k=1:3
for i = 1 : N
for j = 1:4
P_test(i+(k-1)*N,j) = iris(i+(k-1)*50+M,j+1);
end
end
end
[P_p,minp,maxp] = premnmx(P_test);
%P_p=P_test;
P_p=P_p';
Y=sim(net,P_p)
figure
n1=0;n2=0;n3=0;%统计各组分类正确数
for k=1:3
for i=1:N
switch k
case 1
plot(Y(1,i+(k-1)*N),Y(2,i+(k-1)*N),'r+')
if(Y(1,i)>0.5 && Y(2,i)<0.5)
n1=n1+1;
end
case 2
plot(Y(1,i+(k-1)*N),Y(2,i+(k-1)*N),'gs')
if(Y(1,i+N)<0.5 && Y(2,i+N)>0.5)
n2=n2+1;
end
case 3
plot(Y(1,i+(k-1)*N),Y(2,i+(k-1)*N),'b*')
if(Y(1,i+2*N)>0.5 && Y(2,i+2*N)>0.5)
n3=n3+1;
end
end
hold on
end
end
hold on
plot(0:0.001:1,0.5,'--b' )
hold on
plot(0.5,0:0.001:1,'--b')
axis([-0.2 1.2 -0.2 1.2])
xlabel('Y(1)')
ylabel('Y(2)')
title('分类结果')
grid on
评论0