BP神经网络识别性别

function [ data,label ] = getdata( xlsfile )
% [data,label]=getdata('sex.xls')
% 在sex.xls中读入身高和体重
[~,label]=xlsread(xlsfile,'sheet1','B2:B361');
size(label)
[height,~]=xlsread(xlsfile,'sheet1','C2:C361');
[weight,~]=xlsread(xlsfile,'sheet1','D2:D361');
data=[height,weight];
l=zeros(size(label));
for i=1:length(l)
 if strcmp(label{i},'boy')
%     if label{i} == 'boy'
        l(i)=1;
 end
end
label = l;
end

function [ traind,trainl,testd,testl ] = divide( data,label )
% [data,label]=getdata('sex.xls');
% 男女各取30个进行训练
TRAIN_NUM_M =30;
TRAIN_NUM_F=30;
% 男女分开
m_data=data(label==1,:);
f_data=data(label==0,:);
NUM_M=length(m_data);%男生的个数
r=randperm(NUM_M);
traind(1:TRAIN_NUM_M,:) = m_data(r(1:TRAIN_NUM_M),:);
testd(1:NUM_M-TRAIN_NUM_M,:)=m_data(r(TRAIN_NUM_M+1:NUM_M),:);
NUM_F=length(f_data);%女生的个数
r=randperm(NUM_F);
traind(TRAIN_NUM_M+1:TRAIN_NUM_M+TRAIN_NUM_F,:) = f_data(r(1:TRAIN_NUM_F),:);
testd(NUM_M-TRAIN_NUM_M+1:NUM_M-TRAIN_NUM_M+NUM_F-TRAIN_NUM_F,:)=f_data(r(TRAIN_NUM_F+1:NUM_F),:);
trainl=zeros(1,TRAIN_NUM_M+TRAIN_NUM_F);
trainl(1:TRAIN_NUM_M)=1;
testl=zeros(1,NUM_M+NUM_F-TRAIN_NUM_M-TRAIN_NUM_F);
testl(1:NUM_M-TRAIN_NUM_M)=1;
end

clear all;
xlsfile='sex.xls';
[data,label]=getdata(xlsfile);
[traind,trainl,testd,testl]=divide(data,label);
% 设置参数
rng('default')
rng(0)
nTrainNum=60;%60个训练样本
nSampDim=2;%样本是二维的
% 构造网络
net.nIn=2;
net.nHidden=3;%3个隐含节点
net.nOut=1;
w1=2*(rand(net.nHidden,net.nIn)-1/2);
b1=2*(rand(net.nHidden,1)-1/2);
net.w1=[w1,b1];
W=2*(rand(net.nOut,net.nHidden)-1/2);
B=2*(rand(net.nOut,1)-1/2);
net.w2=[W,B];
% 训练数据归一化
mm=mean(traind);
% 均值平移
for i=1:2
    traind_s(:,i)=traind(:,i)-mm(i);
end
% 方差标准化
ml(1)=std(traind_s(:,1));
ml(2)=std(traind_s(:,2));
for i=1:2
    traind_s(:,i)=traind(:,i)/ml(1);
end
% 训练
SampInEx=[traind_s';ones(1,nTrainNum)];
expectedOut=trainl;
eb=0.01;
eta=0.6;
mc=0.8;
maxiter=2000;
iteration=0;
errRec=zeros(1,maxiter);
outRec=zeros(nTrainNum,maxiter);
NET=[];
% 开始迭代
for i=1:maxiter
    hid_input=net.w1*SampInEx;
    hid_out=logsig(hid_input);
   ou_input1=[hid_out;ones(1,nTrainNum)];
   ou_input2=net.w2*ou_input1;
   out_out=logsig(ou_input2);
   outRec(:,i)=out_out';%记录每次迭代输出
   err=expectedOut-out_out;%误差
   sse=sumsqr(err);
   errRec(i)=sse;
   fprintf('第%d次迭代     误差：    %f\n',i,sse);
   iteration=iteration+1;
 %判断是否收敛
 if sse<=eb
 break;
 end
 %误差反向传播
 %隐含层与输出层之间的局部梯度
   delta2=err.*dlogsig(ou_input2,out_out);
 %输入层与隐含层之间的局部梯度
   delta1=net.w2(:,1:end-1)'*delta2.*dlogsig(hid_input,hid_out);
 %权值修改量
   dwex2=delta2*ou_input1';
   dwex1=delta1*SampInEx';
 %修改权值，如果不是第一次修改，则使用动量因子
 if i == 1
       net.w2=net.w2+eta*dwex2;
       net.w1=net.w1+eta*dwex1;
 else
       net.w2=net.w2+(1-mc)*eta*dwex2+mc*dwexold2;
       net.w1=net.w1+(1-mc)*eta*dwex1;+mc*dwexold1;
 end
%    记录上一次的权值修改量
    dwexold2=dwex2;
    dwexold1=dwex1;
end
% 测试数据归一化
for i=1:2
    testd_s(:,i)=testd(:,i)-mm(i);
end
for i=1:2
    testd_s(:,i)=testd(:,i)/mm(i);
end
% 计算测试输出
InEx=[testd_s';ones(1,360-nTrainNum)];
hid_input=net.w1*InEx;
hid_out=logsig(hid_input);
ou_input1=[hid_out;ones(1,360-nTrainNum)];
ou_input2=net.w2*ou_input1;
out_out=logsig(ou_input2);
out_out1=out_out;
%取整
out_out(out_out<0.5)=0;
out_out(out_out>=0.5)=1;
%正确率
rate=sum(out_out==testl)/length(out_out);
% 显示训练样本
train_m=traind(trainl==1,:);
train_m=train_m';
train_f=traind(trainl==0,:);
train_f=train_f';
figure(1)
plot(train_m(1,:),train_m(2,:),'bo');
hold on;
plot(train_f(1,:),train_f(2,:),'r*');