|
#include "BpNet.h"
#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif
/////////////////////////////////////////////////////////////////////////////
// CBpNet
IMPLEMENT_SERIAL( CBpNet, CObject, 1 )
CBpNet::CBpNet()
{initM(MATCOM_VERSION);//启用矩阵运算库
}
CBpNet::~CBpNet()
{exitM();
delete this;
}
/////////////////////////////////////////////////////////////////////////////
// CBpNet message handlers
//创建新网络
void CBpNet::Create(Mm mInputData, Mm mTarget, int iInput, int iHidden, int iOutput)
{ int i,j;
mSampleInput=zeros(mInput.rows(),mInput.cols());
mSampleTarget=zeros(mTarget.rows(),mTarget.cols());
mSampleInput=mInputData;
mSampleTarget=mTarget;
this->iInput=iInput;
this->iHidden=iHidden;
this->iOutput=iOutput;
//创建计算用的单个样本矩阵
mInput=zeros(1,this->iInput);
mHidden=zeros(1,this->iHidden);
mOutput=zeros(1,this->iOutput);
//创建权重矩阵,并赋初值
mWeighti=zeros(this->iInput,this->iHidden);
mWeighto=zeros(this->iHidden,this->iOutput);
//赋初值
for(i=1;iiInput;i++)
for(j=1;jiHidden;j++)
mWeighti.r(i,j)=randab(-1.0,1.0);
for(i=1;iiHidden;i++)
for(j=1;jiOutput;j++)
mWeighto.r(i,j)=randab(-1.0,1.0);
//创建阙值矩阵,并赋值
mThresholdi=zeros(1,this->iHidden);
for(i=1;iiHidden;i++)
mThresholdi.r(i)=randab(-1.0,1.0);
mThresholdo=zeros(1,this->iOutput);
for(i=1;iiOutput;i++)
mThresholdo.r(i)=randab(-1.0,1.0);
//创建权重变化矩阵
mChangei=zeros(this->iInput,this->iHidden);
mChangeo=zeros(this->iHidden,this->iOutput);
mInputNormFactor=zeros(iInput,2);
mTargetNormFactor=zeros(iOutput,2);
//误差矩阵
mOutputDeltas=zeros(iOutput);
mHiddenDeltas=zeros(iHidden);
//学习速率赋值
dblLearnRate1=0.5;
dblLearnRate2=0.5;
dblMomentumFactor=0.95;
m_isOK=false;
m_IsStop=false;
dblMse=1.0e-6;//误差限
dblError=1.0;
lEpochs=0;
}
//根据已有的网络进行预测
Mm CBpNet::simulate(Mm mData)
{int i,j;
Mm mResult;
Mm data=zeros(mData.rows(),mData.cols());
data=mData;
if(mData.cols()!=iInput)
{::MessageBox(NULL,"输入数据变量个数错误!","输入数据变量个数错误!",MB_OK);
return mResult;
}
mResult=zeros(data.rows(),iOutput);
//正规化数据
for(i=1;i for(j=1;j data.r(i,j)=(data.r(i,j)-mInputNormFactor.r(j,1))/(mInputNormFactor.r(j,2)-mInputNormFactor.r(j,1));
//计算
int iSample;
Mm mInputdata,mHiddendata,mOutputdata;
mInputdata=zeros(1,iInput);
mHiddendata=zeros(1,iHidden);
mOutputdata=zeros(1,iOutput);
double sum=0.0;
for(iSample=1;iSample //输入层数据
for(i=1;i mInputdata.r(i)=data.r(iSample,i);
//隐层数据
for(j=1;j sum=0.0;
for(i=1;i sum+=mInputdata.r(i)*mWeighti.r(i,j);
sum-=mThresholdi.r(j);
mHiddendata.r(j)=1.0/(1.0+exp(-sum));
}
//输出数据
for(j=1;j sum=0.0;
for(i=1;i sum+=mHiddendata.r(i)*mWeighto.r(i,j);
sum-=mThresholdo.r(j);
mOutputdata.r(j)=1.0/(1.0+exp(-sum));
}
//转换
for(j=1;j mResult.r(iSample,j)=mOutputdata.r(j)*(mTargetNormFactor.r(j,2)-mTargetNormFactor.r(j,1))+mTargetNormFactor.r(j,1);
}
return (mResult);
}
void CBpNet::LoadBpNet(CString &strNetName)
{CFile file;
if(file.Open(strNetName,CFile::modeRead)==0)
{MessageBox(NULL,"无法打开文件!","错误",MB_OK);
return;
}
else{
CArchive myar(&file,CArchive::load);
Serialize(myar);
myar.Close();
}
file.Close();
}
bool CBpNet::SaveBpNet(CString &strNetName)
{CFile file;
if(strNetName.GetLength()==0)
return(false);
if(file.Open(strNetName,CFile::modeCreate|CFile::modeWrite)==0)
{MessageBox(NULL,"无法创建文件!","错误",MB_OK);
return(false);
}
else{
CArchive myar(&file,CArchive::store);
Serialize(myar);
myar.Close();
}
file.Close();
return(true);
}
//网络学习
void CBpNet::learn()
{ int iSample=1;
double dblTotal;
MSG msg;
if(m_IsStop)
m_IsStop=false;
//数据正规化处理
normalize();
while(dblError>dblMse&&!m_IsStop){
dblTotal=0.0;
for(iSample=1;iSample forward(iSample);
backward(iSample);
dblTotal+=dblErr;//总误差
}
if(dblTotal/dblError>1.04){//动态改变学习速率
dblLearnRate1*=0.7;
dblLearnRate2*=0.7;
}
else{
dblLearnRate1*=1.05;
dblLearnRate2*=1.05;
}
lEpochs++;
dblError=dblTotal;
::PeekMessage(&msg,NULL,0,0,PM_REMOVE);
::DispatchMessage(&msg);
msg.message=-1;
::DispatchMessage(&msg);//这样可以消除屏闪和假死机
}
if(dblError m_isOK=true;
else
m_isOK=false;
}
void CBpNet::stop()
{
m_IsStop=true;
}
double CBpNet::randab(double a, double b)
{ //注意,如果应用矩阵库,头文件matlib.h对rand()函数重新定义,只产生(0,1)
//之间的随机数
return((b-a)*rand()+a);
}
//将数据转化到(0,1)区间
void CBpNet::normalize()
{
int i,j;
//输入数据范围
mInputNormFactor=scope(mSampleInput);
//目标数据范围
mTargetNormFactor=scope(mSampleTarget);
for(i=1;i for(j=1;j mSampleInput.r(i,j)=(mSampleInput.r(i,j)-mInputNormFactor.r(j,1))/(mInputNormFactor.r(j,2)-mInputNormFactor.r(j,1));
for(i=1;i for(j=1;j mSampleTarget.r(i,j)=(mSampleTarget.r(i,j)-mTargetNormFactor.r(j,1))/(mTargetNormFactor.r(j,2)-mTargetNormFactor.r(j,1));
}
//前向计算
void CBpNet::forward(int iSample)
{//根据第iSample个样本,前向计算
if(iSamplemSampleInput.rows()){
MessageBox(NULL,"无此样本数据:索引出界!","无此样本数据:索引出界!",MB_OK);
return;
}
int i,j;
double sum=0.0;
//输入层数据
for(i=1;i mInput.r(i)=mSampleInput.r(iSample,i);
//隐层数据
for(j=1;j sum=0.0;
for(i=1;i sum+=mInput.r(i)*mWeighti.r(i,j);
sum-=mThresholdi.r(j);
mHidden.r(j)=1.0/(1.0+exp(-sum));
}
//输出数据
for(j=1;j sum=0.0;
for(i=1;i sum+=mHidden.r(i)*mWeighto.r(i,j);
sum-=mThresholdo.r(j);
mOutput.r(j)=1.0/(1.0+exp(-sum));
}
}
//后向反馈
void CBpNet::backward(int iSample)
{
if(iSamplemSampleInput.rows()){
MessageBox(NULL,"无此样本数据:索引出界!","无此样本数据:索引出界!",MB_OK);
return;
}
int i,j;
//输出误差
for(i=1;i mOutputDeltas.r(i)=mOutput.r(i)*(1-mOutput.r(i))*(mSampleTarget.r(iSample,i)-mOutput.r(i));
//隐层误差
double sum=0.0;
for(j=1;j sum=0.0;
for(i=1;i sum+=mOutputDeltas.r(i)*mWeighto.r(j,i);
mHiddenDeltas.r(j)=mHidden.r(j)*(1-mHidden.r(j))*sum;
}
//更新隐层-输出权重
double dblChange;
for(j=1;j for(i=1;i dblChange=mOutputDeltas.r(i)*mHidden.r(j);
mWeighto.r(j,i)=mWeighto.r(j,i)+dblLearnRate2*dblChange+dblMomentumFactor*mChangeo.r(j,i);
mChangeo.r(j,i)=dblChange;
}
//更新输入-隐层权重
for(i=1;i for(j=1;j dblChange=mHiddenDeltas.r(j)*mInput.r(i);
mWeighti.r(i,j)=mWeighti.r(i,j)+dblLearnRate1*dblChange+dblMomentumFactor*mChangei.r(i,j);
mChangei.r(i,j)=dblChange;
}
//修改阙值
for(j=1;j mThresholdo.r(j)-=dblLearnRate2*mOutputDeltas.r(j);
for(i=1;i mThresholdi.r(i)-=dblLearnRate1*mHiddenDeltas.r(i);
//计算误差
dblErr=0.0;
for(i=1;i dblErr+=0.5*(mSampleTarget.r(iSample,i)-mOutput.r(i))*(mSampleTarget.r(iSample,i)-mOutput.r(i));
}
//求数据列的范围
Mm CBpNet::scope(Mm mData)
{Mm mScope;
mScope=zeros(mData.cols(),2);
double min,max;
for(int i=1;i min=max=mData.r(1,i);
for(int j=1;j if(mData.r(j,i)>=max)
max=mData.r(j,i);
if(mData.r(j,i) min=mData.r(j,i);
}
if(min==max)
min=0.0;
mScope.r(i,1)=min;
mScope.r(i,2)=max;
}
return(mScope);
}
//显示矩阵数据,方便调试
void CBpNet::display(Mm data)
{CString strData,strTemp;
int i=1,j=1;
for(i=1;i for(j=1;j strTemp.Format("%.3f ",data.r(i,j));
strData+=strTemp;
}
strData=strData+"\r\n";
}
::MessageBox(NULL,strData,"",MB_OK);
}
void CBpNet::Serialize(CArchive &ar)
{CObject::Serialize(ar);
/////////////////////////////////////
if(ar.IsStoring()){
int i,j;
double dblData;
CString strTemp="Bp";
ar //纪录神经元个数
ar //纪录权值
for(i=1;i for(j=1;j dblData=mWeighti.r(i,j);
ar }
for(i=1;i for(j=1;j dblData=mWeighto.r(i,j);
ar }
//记录权值变化
for(j=1;j for(i=1;i ar
//输入-隐层权重变化
for(i=1;i for(j=1;j ar
//纪录阙值
for(i=1;i dblData=mThresholdi.r(i);
ar }
for(i=1;i dblData=mThresholdo.r(i);
ar }
//纪录输入输出的极值
for(i=1;i dblData=mInputNormFactor.r(i,1);
ar dblData=mInputNormFactor.r(i,2);
ar }
for(i=1;i {dblData=mTargetNormFactor.r(i,1);
ar dblData=mTargetNormFactor.r(i,2);
ar }
//误差范围
ar //学习速率
ar
}
else{
int i,j;
CString strTemp="";
double dblTemp;
ar>>strTemp;//读入标志
//读入神经元个数
ar>>iInput>>iHidden>>iOutput;
mChangei=zeros(iInput,iHidden);
mChangeo=zeros(iHidden,iOutput);
mWeighti=zeros(iInput,iHidden);
mWeighto=zeros(iHidden,iOutput);
//读入权值
for(i=1;i for(j=1;j { ar>>dblTemp;
mWeighti.r(i,j)=dblTemp;
}
for(i=1;i for(j=1;j { ar>>dblTemp;
mWeighto.r(i,j)=dblTemp;
}
//读入权值变化
for(j=1;j for(i=1;i ar>>mChangeo.r(j,i);
//输入-隐层权重
for(i=1;i for(j=1;j ar>>mChangei.r(i,j);
//读入阙值
mThresholdi=zeros(1,iHidden);
for(i=1;i {ar>>dblTemp;
mThresholdi.r(i)=dblTemp;
}
mThresholdo=zeros(1,iOutput);
for(i=1;i {ar>>dblTemp;
mThresholdo.r(i)=dblTemp;
}
//读入输入输出的极值
mInputNormFactor=zeros(iInput,2);
for(i=1;i ar>>dblTemp;
mInputNormFactor.r(i,1)=dblTemp; //极小值
ar>>dblTemp;
mInputNormFactor.r(i,2)=dblTemp; //极大值
}
mTargetNormFactor=zeros(iOutput,2);
for(i=1;i {ar>>dblTemp;
mTargetNormFactor.r(i,1)=dblTemp; //输出数据极小值
ar>>dblTemp;
mTargetNormFactor.r(i,2)=dblTemp;
}
//读入误差范围
ar>>dblMse;
//读入学习速率
ar>>dblLearnRate1>>dblLearnRate2;
//创建计算用的单个样本矩阵
mInput=zeros(1,iInput);
mHidden=zeros(1,iHidden);
mOutput=zeros(1,iOutput);
//误差矩阵
mOutputDeltas=zeros(iOutput);
mHiddenDeltas=zeros(iHidden);
}
}
//如果不是新网络,比如从文件恢复的网络,调用此函数构建学习样本
void CBpNet::LoadPattern(Mm mIn, Mm mOut)
{ if(mIn.cols()!=iInput||mOut.cols()!=iOutput){
::MessageBox( NULL,"学习样本格式错误!","错误",MB_OK);
return;
}
mSampleInput=zeros(mIn.rows(),mIn.cols());
mSampleTarget=zeros(mOut.rows(),mOut.cols());
mSampleInput=mIn;
mSampleTarget=mOut;
m_isOK=false;
m_IsStop=false;
lEpochs=0;
dblMomentumFactor=0.95;
dblError=1.0;
} |
阿莫论坛20周年了!感谢大家的支持与爱护!!
如果天空是黑暗的,那就摸黑生存;
如果发出声音是危险的,那就保持沉默;
如果自觉无力发光,那就蜷伏于牆角。
但是,不要习惯了黑暗就为黑暗辩护;
也不要为自己的苟且而得意;
不要嘲讽那些比自己更勇敢的人。
我们可以卑微如尘土,但不可扭曲如蛆虫。
|