twirls/GMM/main.cpp

220 lines
6.5 KiB
C++
Raw Normal View History

/*********************************************************************************************
Description: <EFBFBD>ø<EFBFBD>˹<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>ͽ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ߵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
Copyright : All right reserved by ZheYuan.BJ
Author : Zhang Zhonghai
Date : 2023/09/12
***********************************************************************************************/
#include <iostream>
#include <iomanip>
#include <fstream>
#include <sstream>
#include <algorithm>
#include <random>
#include <unordered_map>
#include <omp.h>
#include <time.h>
#include <string>
#include <vector>
#include <queue>
#include <filesystem>
#include <thread>
#ifdef _WIN32
#include <io.h>
#include <process.h>
#define F_OK 0
#else
#include <unistd.h>
#endif
#include <mat.h>
#include "gmm.h"
// #include "CommonLib/thread_pool.h"
#include "CommonLib/matlab_io.h"
#include "CommonLib/kthread.h"
using namespace std;
using std::cout;
using std::vector;
namespace fs = std::filesystem;
/* <20><>x<EFBFBD><78><EFBFBD><EFBFBD><EFBFBD>ŵ<EFBFBD><C5B5><EFBFBD><EFBFBD><EFBFBD>ΪbinWidth<74><68>С<EFBFBD><D0A1>Ͱ<EFBFBD>У<EFBFBD><D0A3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>matlab<61><62>hist*/
void PutXtoBin(double* x, int xSize, double binWidth, vector<double>& vXBin, vector<double>& vYBin) {
double maxX = 0.0;
for (int i = 0; i < xSize; ++i) {
if (maxX < x[i]) maxX = x[i];
}
int binSize = (int)((maxX + binWidth / 2) / binWidth + 1);
double binMaxVal = (binSize - 1) * binWidth;
if (binMaxVal > maxX) { // ȷ<><C8B7><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><D2BB>bin<69><6E><EFBFBD><EFBFBD><EFBFBD><EFBFBD>maxX<78><58><EFBFBD><EFBFBD><EFBFBD>Ҳ<EFBFBD>С<EFBFBD><D0A1>maxX-binWidth
binSize -= 1;
}
vXBin.resize(xSize);
vYBin.resize(binSize);
for (int i = 0; i < binSize; ++i) vYBin[i] = 0;
for (int i = 0; i < xSize; ++i) {
int binIdx = (int)((x[i] + binWidth / 2) / binWidth);
if (binIdx >= binSize) binIdx = binSize - 1;
vYBin[binIdx] += 1;
}
// <20><><EFBFBD><EFBFBD>С˳<D0A1><CBB3><EFBFBD><EFBFBD><EFBFBD>޸ĺ<DEB8><C4BA><EFBFBD>x<EFBFBD><78>ֵ<EFBFBD><EFBFBD><E6B4A2>vXBin<69>У<EFBFBD><D0A3><EFBFBD><EFBFBD><EFBFBD>˳<EFBFBD><CBB3><EFBFBD><EFBFBD>ͬ<EFBFBD><CDAC>ѵ<EFBFBD><D1B5><EFBFBD><EFBFBD><EFBFBD>ĸ<EFBFBD>˹<EFBFBD><CBB9><EFBFBD><EFBFBD>ģ<EFBFBD>Ͳ<EFBFBD><CDB2><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һЩ<D2BB><D0A9>ͬ<EFBFBD><CDAC>
int xIdx = 0;
for (int i = 0; i < binSize; ++i) {
for (int j = 0; j < vYBin[i]; ++j) {
vXBin[xIdx++] = i * binWidth;
}
}
}
/* <20><><EFBFBD><EFBFBD>׼<EFBFBD><D7BC>˹ģ<CBB9><C4A3>ѵ<EFBFBD><D1B5><EFBFBD><EFBFBD><EFBFBD>IJ<EFBFBD><C4B2><EFBFBD>ת<EFBFBD><D7AA><EFBFBD><EFBFBD><EFBFBD>Զ<EFBFBD><D4B6><EFBFBD><EFBFBD><EFBFBD>ϵ<EFBFBD><CFB5>, <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϻ<EFBFBD><CFBA><EFBFBD><59><D6B5><EFBFBD><EFBFBD> */
struct cmpFunc {
bool operator()(const pair<double, double>& a, const pair<double, double>& b) { return a.first < b.first; }
};
void GMMToFactorEY(GMM& gmm, double binWidth, vector<double> &vYBin, vector<double>& vFactor, vector<double>& vEY) {
/* <20><>Ҫ<EFBFBD><D2AA><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ߵ<EFBFBD>Ȩ<EFBFBD>أ<EFBFBD><D8A3><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ϸ<EFBFBD>˹<EFBFBD><CBB9><EFBFBD>ߣ<EFBFBD><DFA3><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ø<EFBFBD><C3B8><EFBFBD><EFBFBD>ܶ<EFBFBD> */
double zoomFactorSum = 0.0;
vEY.resize(vYBin.size());
int topM = (int)(vYBin.size() / 4);
if (topM < 1) topM = 1;
/* <20>ö<EFBFBD><C3B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ķ<EFBFBD>ʽȡǰtopM<70><4D><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ, <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ų<EFBFBD><C5B2><EFBFBD>*/
priority_queue<pair<double, double>, vector<pair<double, double> >, cmpFunc> pqTopM;
for (int i = 0; i < vYBin.size(); ++i) {
double xVal = i * binWidth;
double probVal = gmm.GetProbability(&xVal);
vEY[i] = probVal;
pqTopM.push(make_pair(vYBin[i], probVal));
}
for (int i = 0; i < topM; ++i) {
pair<double, double> topEle = pqTopM.top();
pqTopM.pop();
zoomFactorSum += topEle.first / topEle.second;
}
double zoomFactor = zoomFactorSum / topM;
for (int i = 0; i < vEY.size(); ++i) {
vEY[i] *= zoomFactor;
}
vFactor.clear();
vFactor.push_back(zoomFactor * gmm.Prior(0) / sqrt(2 * M_PI * *gmm.Variance(0)));
vFactor.push_back(*gmm.Mean(0));
vFactor.push_back(sqrt(2 * *gmm.Variance(0)));
vFactor.push_back(zoomFactor * gmm.Prior(1) / sqrt(2 * M_PI * *gmm.Variance(1)));
vFactor.push_back(*gmm.Mean(1));
vFactor.push_back(sqrt(2 * *gmm.Variance(1)));
}
/* <20><><EFBFBD><EFBFBD>ƽ<EFBFBD><C6BD><EFBFBD><EFBFBD> */
template <typename T>
T Average(vector<T>& vVal) {
T sumVal = T(0);
for (int i = 0; i < vVal.size(); ++i) {
sumVal += vVal[i];
}
return sumVal / vVal.size();
}
/* <20><><EFBFBD><EFBFBD>ƽ<EFBFBD><C6BD><EFBFBD>ľ<EFBFBD>ֵ */
template <typename T>
T SquareAverage(vector<T>& vVal) {
vector<T> vSquare(vVal.size());
for (int i = 0; i < vVal.size(); ++i) {
vSquare[i] = vVal[i] * vVal[i];
}
return Average(vSquare);
}
/* <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>x<EFBFBD><78>y<EFBFBD><79><EFBFBD><EFBFBD><EFBFBD>ؾ<EFBFBD><D8BE><EFBFBD>, <20><><EFBFBD><EFBFBD>ά<EFBFBD>ȱ<EFBFBD><C8B1><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>*/
double CorrelationDistance(vector<double>& vX, vector<double>& vY) {
vector<double> vXY(vX.size());
for (int i = 0; i < vXY.size(); ++i) {
vXY[i] = vX[i] * vY[i];
}
double uv = Average(vXY);
double uu = SquareAverage(vX);
double vv = SquareAverage(vY);
double dist = 1.0 - uv / sqrt(uu * vv);
return abs(dist);
}
/* <20><><EFBFBD><EFBFBD>һ<EFBFBD><D2BB>֪ʶ<D6AA><CAB6><EFBFBD><EFBFBD> */
struct ThreadParam {
fs::path matFilePath;
fs::path outFilePath;
};
//void ThreadProcessData(vector<ThreadParam>& vTP, long idx, int tid) {
void ThreadProcessData(ThreadParam& param) {
//const ThreadParam& param = vTP[idx];
const fs::path& matFilePath = param.matFilePath;
const fs::path& outFilePath = param.outFilePath;
double* hs = nullptr;
int rowNum = 0;
int colNum = 0;
hs = ReadMtxDouble(matFilePath.string(), "hs", &rowNum, &colNum);
vector<double>vXBin;
vector<double>vYBin;
vector<double>vEY;
vector<double>vFactor;
/* <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ݣ<EFBFBD><DDA3><EFBFBD><EFBFBD><EFBFBD>mat<61>ļ<EFBFBD> */
vector<double>vDist(rowNum);
vector<double>vFactorAll;
for (int i = 0; i < rowNum; ++i) {
PutXtoBin(hs + i * colNum, colNum, 0.2, vXBin, vYBin);
GMM gmm(1, 2); // 1ά<31><CEAC> 2<><32><EFBFBD><EFBFBD>˹ģ<CBB9><C4A3>
gmm.Train(vXBin.data(), (int)vXBin.size());
GMMToFactorEY(gmm, 0.2, vYBin, vFactor, vEY);
vDist[i] = CorrelationDistance(vYBin, vEY);
vFactorAll.insert(vFactorAll.end(), vFactor.begin(), vFactor.end());
}
/* д<><D0B4>matlab<61>ļ<EFBFBD> */
MATFile* pMatFile = matOpen(outFilePath.string().c_str(), "w");
SaveMtxDouble(vFactorAll.data(), pMatFile, "factor", rowNum, 6);
SaveMtxDouble(vDist.data(), pMatFile, "correlation", rowNum, 1);
matClose(pMatFile);
delete[] hs;
}
2023-09-18 02:38:24 +08:00
/* <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> */
int main(int argc, const char** argv) {
if (argc < 4) {
cout << "This program take at least 3 arguments(1.parrent Dir; 2. mat file suffix; 3. out mat filename; [4. thread number])!" << endl;
return 1;
}
string parrentDir(argv[1]); // ֪ʶ<D6AA><CAB6><EFBFBD><EFBFBD><EFBFBD>ĸ<EFBFBD>Ŀ¼<C4BF><C2BC><EFBFBD><EFBFBD>
string hsMatSuffix(argv[2]); // hs<68><73><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ӧ<EFBFBD><D3A6>mat<61>ļ<EFBFBD><C4BC>ĺ<EFBFBD>׺<EFBFBD><D7BA><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD>ļ<EFBFBD><C4BC><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ļ<EFBFBD><C4BC><EFBFBD><EFBFBD><EFBFBD>׺<EFBFBD><D7BA><EFBFBD><EFBFBD><EFBFBD>뱣֤Ψһ<CEA8><D2BB>
fs::path outFileName(argv[3]);
int numThread = 1;
if (argc >= 4) numThread = atoi(argv[4]);
if (numThread < 1) numThread = 1;
//ThreadPool thPool(numThread);
clock_t begin, finish;
begin = clock();
vector<ThreadParam> vTP;
2023-09-18 02:38:24 +08:00
/* <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>е<EFBFBD>֪ʶ<D6AA><CAB6><EFBFBD><EFBFBD>Ŀ¼<C4BF><C2BC><EFBFBD><EFBFBD>һ<EFBFBD><D2BB><EFBFBD>д<EFBFBD><D0B4><EFBFBD> */
for (auto& childDir : fs::directory_iterator(parrentDir)) {
fs::path outFilePath = childDir / outFileName;
for (auto& file : fs::directory_iterator(childDir)) {
const string& fileName = file.path().filename().string();
auto rPos = fileName.rfind(hsMatSuffix);
if (rPos != string::npos && fileName.size() - rPos == hsMatSuffix.size()) {
//ThreadParam tParam = { file, outFilePath };
//thPool.enqueue(ThreadProcessData, tParam);
vTP.push_back({ file, outFilePath });
}
}
}
kt_for(numThread, ThreadProcessData, vTP);
//thPool.~ThreadPool();
finish = clock();
cout << "GMM Total time: " << (double)(finish - begin) / CLOCKS_PER_SEC << " s" << endl;
return 0;
}