BP 神经网络 C++实现

来源：互联网发布：sqlserver怎么用编辑：程序博客网时间：2024/05/19 12:26

BP.h

#ifndef _BP_H_#define _BP_H_ #include <vector> #define LAYER    3        //三层神经网络#define NUM      10       //每层的最多节点数 #define A        30.0#define B        10.0     //A和B是S型函数的参数#define ITERS    1000     //最大训练次数#define ETA_W    0.0035   //权值调整率#define ETA_B    0.001    //阀值调整率#define ERROR    0.002    //单个样本允许的误差#define ACCU     0.005    //每次迭代允许的误差 #define Type double#define Vector std::vector struct Data{    Vector<Type> x;       //输入数据    Vector<Type> y;       //输出数据}; class BP{ public:     void GetData(const Vector<Data>);    void Train();    Vector<Type> ForeCast(const Vector<Type>); private:     void InitNetWork();         //初始化网络    void GetNums();             //获取输入、输出和隐含层节点数    void ForwardTransfer();     //正向传播子过程    void ReverseTransfer(int);  //逆向传播子过程    void CalcDelta(int);        //计算w和b的调整量    void UpdateNetWork();       //更新权值和阀值    Type GetError(int);         //计算单个样本的误差    Type GetAccu();             //计算所有样本的精度    Type Sigmoid(const Type);   //计算Sigmoid的值 private:    int in_num;                 //输入层节点数    int ou_num;                 //输出层节点数    int hd_num;                 //隐含层节点数     Vector<Data> data;          //输入输出数据     Type w[LAYER][NUM][NUM];    //BP网络的权值    Type b[LAYER][NUM];         //BP网络节点的阀值         Type x[LAYER][NUM];         //每个神经元的值经S型函数转化后的输出值，输入层就为原值    Type d[LAYER][NUM];         //记录delta学习规则中delta的值}; #endif  //_BP_H_

BP.CPP

#include <string.h>#include <stdio.h>#include <math.h>#include <assert.h>#include "BP.h"//获取训练所有样本数据void BP::GetData(const Vector<Data> _data){data = _data;}//开始进行训练void BP::Train(){printf("Begin to train BP NetWork!\n");GetNums();InitNetWork();    int num = data.size();for(int iter = 0; iter <= ITERS; iter++){for(int cnt = 0; cnt < num; cnt++){//第一层输入节点赋值for(int i = 0; i < in_num; i++)x[0][i] = data.at(cnt).x[i];while(1){ForwardTransfer();                     if(GetError(cnt) < ERROR)    //如果误差比较小，则针对单个样本跳出循环break;                ReverseTransfer(cnt);  }}printf("This is the %d th trainning NetWork !\n", iter);Type accu = GetAccu();printf("All Samples Accuracy is %lf\n", accu);if(accu < ACCU) break;}printf("The BP NetWork train End!\n");}//根据训练好的网络来预测输出值Vector<Type> BP::ForeCast(const Vector<Type> data){int n = data.size();assert(n == in_num);    for(int i = 0; i < in_num; i++)x[0][i] = data[i];    ForwardTransfer();Vector<Type> v;for(int i = 0; i < ou_num; i++)v.push_back(x[2][i]);return v;}//获取网络节点数void BP::GetNums(){in_num = data[0].x.size();                         //获取输入层节点数ou_num = data[0].y.size();                         //获取输出层节点数    hd_num = (int)sqrt((in_num + ou_num) * 1.0) + 5;   //获取隐含层节点数if(hd_num > NUM) hd_num = NUM;                     //隐含层数目不能超过最大设置}//初始化网络void BP::InitNetWork(){    memset(w, 0, sizeof(w));      //初始化权值和阀值为0，也可以初始化随机值memset(b, 0, sizeof(b));}//工作信号正向传递子过程void BP::ForwardTransfer(){//计算隐含层各个节点的输出值for(int j = 0; j < hd_num; j++){Type t = 0;for(int i = 0; i < in_num; i++)t += w[1][i][j] * x[0][i];t += b[1][j];x[1][j] = Sigmoid(t);}//计算输出层各节点的输出值for(int j = 0; j < ou_num; j++){Type t = 0;for(int i = 0; i < hd_num; i++)t += w[2][i][j] * x[1][i];t += b[2][j];x[2][j] = Sigmoid(t);}}//计算单个样本的误差Type BP::GetError(int cnt){Type ans = 0;for(int i = 0; i < ou_num; i++)ans += 0.5 * (x[2][i] - data.at(cnt).y[i]) * (x[2][i] - data.at(cnt).y[i]);return ans;}//误差信号反向传递子过程void BP::ReverseTransfer(int cnt){CalcDelta(cnt);   UpdateNetWork();}//计算所有样本的精度Type BP::GetAccu(){Type ans = 0;int num = data.size();for(int i = 0; i < num; i++){int m = data.at(i).x.size();for(int j = 0; j < m; j++)x[0][j] = data.at(i).x[j];        ForwardTransfer();int n = data.at(i).y.size();for(int j = 0; j < n; j++)            ans += 0.5 * (x[2][j] - data.at(i).y[j]) * (x[2][j] - data.at(i).y[j]);}return ans / num;}//计算调整量void BP::CalcDelta(int cnt){//计算输出层的delta值for(int i = 0; i < ou_num; i++)d[2][i] = (x[2][i] - data.at(cnt).y[i]) * x[2][i] * (A - x[2][i]) / (A * B);//计算隐含层的delta值for(int i = 0; i < hd_num; i++){Type t = 0;for(int j = 0; j < ou_num; j++)t += w[2][i][j] * d[2][j];d[1][i] = t * x[1][i] * (A - x[1][i]) / (A * B);}}//根据计算出的调整量对BP网络进行调整void BP::UpdateNetWork(){//隐含层和输出层之间权值和阀值调整for(int i = 0; i < hd_num; i++){for(int j = 0; j < ou_num; j++)w[2][i][j] -= ETA_W * d[2][j] * x[1][i]; }for(int i = 0; i < ou_num; i++)b[2][i] -= ETA_B * d[2][i];//输入层和隐含层之间权值和阀值调整for(int i = 0; i < in_num; i++){for(int j = 0; j < hd_num; j++)w[1][i][j] -= ETA_W * d[1][j] * x[0][i];}for(int i = 0; i < hd_num; i++)b[1][i] -= ETA_B * d[1][i];}//计算Sigmoid函数的值Type BP::Sigmoid(const Type x){return A / (1 + exp(-x / B));}

Test.CPP

#include <iostream>#include <string.h>#include <stdio.h> #include "BP.h" using namespace std; double sample[41][4]= {     {0,0,0,0},     {5,1,4,19.020},     {5,3,3,14.150},     {5,5,2,14.360},     {5,3,3,14.150},     {5,3,2,15.390},     {5,3,2,15.390},     {5,5,1,19.680},     {5,1,2,21.060},     {5,3,3,14.150},     {5,5,4,12.680},     {5,5,2,14.360},     {5,1,3,19.610},     {5,3,4,13.650},     {5,5,5,12.430},     {5,1,4,19.020},     {5,1,4,19.020},     {5,3,5,13.390},     {5,5,4,12.680},     {5,1,3,19.610},     {5,3,2,15.390},     {1,3,1,11.110},     {1,5,2,6.521},     {1,1,3,10.190},     {1,3,4,6.043},     {1,5,5,5.242},     {1,5,3,5.724},     {1,1,4,9.766},     {1,3,5,5.870},     {1,5,4,5.406},     {1,1,3,10.190},     {1,1,5,9.545},     {1,3,4,6.043},     {1,5,3,5.724},     {1,1,2,11.250},     {1,3,1,11.110},     {1,3,3,6.380},     {1,5,2,6.521},     {1,1,1,16.000},     {1,3,2,7.219},     {1,5,3,5.724} };  int main(){    Vector<Data> data;    for(int i = 0; i < 41; i++)    {        Data t;        for(int j = 0; j < 3; j++)            t.x.push_back(sample[i][j]);        t.y.push_back(sample[i][3]);        data.push_back(t);    }    BP *bp = new BP();    bp->GetData(data);    bp->Train();     while(1)    {        Vector<Type> in;        for(int i = 0; i < 3; i++)        {            Type v;            scanf("%lf", &v);            in.push_back(v);        }        Vector<Type> ou;        ou = bp->ForeCast(in);        printf("%lf\n", ou[0]);    }    return 0;}

阅读全文

0 0