bool SvmClassifier::Save(const string &model_name) const
{
// Save the normalization parameters and penalty coefficient
vector<float> c(1, c_);
if (!SaveMat(model_name + "_normA", normA_, c))
{
return false;
}
if (!SaveMat(model_name + "_normB", normB_))
{
return false;
}
// Save the SVM model
if (svm_model_ == NULL)
{
return false;
}
if (svm_save_model((model_name + FILE_EXT).c_str(),
svm_model_) != 0)
{
return false;
}
return true;
}
int main(int argc, char **argv)
{
char input_file_name[1024];
char model_file_name[1024];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"Error: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
do_cross_validation();
}
else
{
model = svm_train(&prob,¶m);
svm_save_model(model_file_name,model);
svm_destroy_model(model);
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
return 0;
}
void svm_train_and_test(const std::vector<feature_t> &feats, const std::vector<int> &labels, const char * model_file, std::ofstream &ofs)
{
svm_model * model;
if(_access(model_file, 0) == -1)
{
auto param = svm_fill_parameter();
auto prob = svm_fill_problem(feats, labels);
model = svm_train(prob, param);
svm_save_model(model_file, model);
auto acc = svm_test_acc(prob, model);
svm_destroy_param(param);
svm_free_problem(prob);
std::cout<<model_file<<" acc: "<<acc*100<<std::endl;
ofs<<model_file<<" acc: "<<acc*100<<std::endl;
}
else
{
model = svm_load_model(model_file);
auto acc = svm_test_acc(feats, labels, model);
std::cout<<model_file<<" acc: "<<acc*100<<std::endl;
ofs<<model_file<<" acc: "<<acc*100<<std::endl;
}
//free
svm_free_and_destroy_model(&model);
}
static void cmd_train(const char *modelfile, double cParam, double gParam,
WindowFile &trainA, WindowFile &trainB)
{
SVMProblem problem(trainA, trainB);
SVMParam param(cParam, gParam, true);
svm_model *model = svm_train(&problem, ¶m);
svm_save_model(modelfile, model);
svm_free_and_destroy_model(&model);
}
int jpcnn_save_predictor(const char* filename, void* predictorHandle) {
SPredictorInfo* predictorInfo = (SPredictorInfo*)(predictorHandle);
struct svm_model* model = predictorInfo->model;
const int saveResult = svm_save_model(filename, model);
if (saveResult != 0) {
fprintf(stderr, "Couldn't save libsvm model file to '%s'\n", filename);
return 0;
}
return 1;
}
double svm_train_and_test(const std::vector<feature_t> &train_feats, const std::vector<int> &train_labels, const std::vector<feature_t> &test_feats, const std::vector<int> &test_labels, const char * model_file)
{
svm_model * model;
auto param = svm_fill_parameter();
auto prob = svm_fill_problem(train_feats, train_labels);
model = svm_train(prob, param);
svm_save_model(model_file, model);
auto acc = svm_test_acc(test_feats, test_labels, model);
svm_destroy_param(param);
svm_free_problem(prob);
//free
svm_free_and_destroy_model(&model);
return acc;
}
void mexFunction(int nlhs, mxArray *plhs[], int nrhs, const mxArray *prhs[])
{
struct svm_model *model;
char *filename;
const char *error_msg;
int status;
/* check input */
if(nrhs != 2) {
mexPrintf("Usage: svm_savemodel(model, 'filename');\n");
fake_answer(plhs);
return;
}
if(!mxIsStruct(prhs[0])) {
mexPrintf("model file should be a struct array\n");
fake_answer(plhs);
return;
}
if(!mxIsChar(prhs[1]) || mxGetM(prhs[1])!=1) {
mexPrintf("filename should be given as char(s)\n");
fake_answer(plhs);
return;
}
/* convert MATLAB struct to C struct */
model = matlab_matrix_to_model(prhs[0], &error_msg);
if(model == NULL) {
mexPrintf("Error: can't read model: %s\n", error_msg);
fake_answer(plhs);
return;
}
/* get filename */
filename = mxArrayToString(prhs[1]);
/* save model to file */
status = svm_save_model(filename,model);
if (status != 0) {
mexWarnMsgTxt("Error occured while writing to file.");
}
/* destroy model */
svm_free_and_destroy_model(&model);
mxFree(filename);
/* return status value (0: success, -1: failure) */
plhs[0] = mxCreateDoubleScalar(status);
return;
}
int main(int argc, char **argv)
{
char input_file_name[1024];
char model_file_name[1024];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
#ifdef CUSTOM_SOLVER
if (param.svm_type == ONE_CLASS)
{
param.strong_footlier_indexes = filter_strong_footliers(input_file_name);
}
#endif
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"Error: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
if(param.svm_type == R2 || param.svm_type == R2q)
fprintf(stderr, "\"R^2\" cannot do cross validation.\n");
else
do_cross_validation();
}
else
{
model = svm_train(&prob,¶m);
if(param.svm_type == R2 || param.svm_type == R2q)
fprintf(stderr, "\"R^2\" does not generate model.\n");
else if(svm_save_model(model_file_name,model))
{
fprintf(stderr, "can't save model to file %s\n", model_file_name);
exit(1);
}
svm_free_and_destroy_model(&model);
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
free(line);
return 0;
}
static VALUE cModel_save(VALUE obj, VALUE filename)
{
const struct svm_model *model;
const char *path;
int rc;
Data_Get_Struct(obj, struct svm_model, model);
path = StringValueCStr(filename);
if(rc = svm_save_model(path, model)) {
rb_raise(rb_eStandardError, "Error on saving model, code: %i", rc);
}
return Qnil;
}
int main(int argc, char **argv)
{
char input_file_name[1024];
char model_file_name[1024];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"ERROR: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
if (nr_fold <= 10)
{
do_cross_validation();
}
else
{
double cv;
nr_fold = nr_fold - 10;
cv = binary_class_cross_validation(&prob, ¶m, nr_fold);
printf("Cross Validation = %g%%\n",100.0*cv);
}
}
else
{
model = svm_train(&prob,¶m);
if(svm_save_model(model_file_name,model))
{
fprintf(stderr, "can't save model to file %s\n", model_file_name);
exit(1);
}
svm_free_and_destroy_model(&model);
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
free(line);
return 0;
}
bool CmySvmArth::Train( char* path)
{
const char *error_msg;
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"ERROR: %s\n",error_msg);
free(prob.y);
free(prob.x);
free(x_space);
free(line);
line = NULL;
x_space = NULL;
return false;
}
if(cross_validation)
{
do_cross_validation();
}
else
{
model = svm_train(&prob,¶m);
if(path!=NULL&&svm_save_model(path,model))
{
fprintf(stderr, "can't save model to file %s\n", path);
return false;
}
if(path!=NULL)
{
svm_free_and_destroy_model(&model);
}
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
if(path!=NULL)
{
free(x_space);
free(line);
line = NULL;
x_space = NULL;
}
return true;
}
int SVMTrainModel::train(double &RecRate, std::vector<int> &ConfusionTable)
{
if((!have_input_file_name) || (!have_model_file_name))
{
fprintf(stderr,"ERROR: Set Input and Model files first!\n");
exit(1);
}
const char *error_msg;
readProblem(input_file_name);
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"ERROR: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
do_cross_validation(RecRate,ConfusionTable);
}
else
{
model = svm_train(&prob,¶m);
if(svm_save_model(model_file_name,model))
{
fprintf(stderr, "can't save model to file %s\n", model_file_name);
exit(1);
}
svm_free_and_destroy_model(&model);
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
free(line);
return 0;
}
int main(int argc, char **argv)
{
char input_file_name[1024];
char model_file_name[1024];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
fprintf(stderr,"ERROR: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
do_cross_validation();
}
else
{
model = svm_train(&prob,¶m);
if(svm_save_model(model_file_name,model))
{
fprintf(stderr, "can't save model to file %s\n", model_file_name);
exit(1);
}
svm_free_and_destroy_model(&model);
}
svm_destroy_param(¶m);
free(prob.y);
#ifdef _DENSE_REP
for (int i = 0; i < prob.l; ++i)
free((prob.x+i)->values);
#else
free(x_space);
#endif
free(prob.x);
free(line);
return 0;
}
int savemodel(const char *filename, const mxArray *matlab_struct)
{
const char *error_msg;
struct svm_model* model;
int result;
model = matlab_matrix_to_model(matlab_struct, &error_msg);
if (model == NULL)
{
mexPrintf("Error: can't read model: %s\n", error_msg);
}
result = svm_save_model(filename, model);
if( result != 0 )
{
mexPrintf("Error: can't write model to file!\n");
}
return result;
}
void SVM::train(char* pInputSampleFileName, char* OutputModelFilename, double &dRetTrainError, double &dRetCrossValError)
{
struct svm_parameter strSvmParameters;
struct svm_problem strSvmProblem;
struct svm_model *pstrSvmModel;
const char *error_msg;
double dCrossValError = -1;
double dTrainError = -1;
//set parameters
this->setParameters(strSvmParameters);
//read sample file
this->read_problem(pInputSampleFileName, strSvmProblem, strSvmParameters);
//check parameters
error_msg = svm_check_parameter(&strSvmProblem, &strSvmParameters);
//train model
pstrSvmModel = svm_train(&strSvmProblem, &strSvmParameters);
//do cross validation check
dCrossValError = this->crossValidationSamples(strSvmProblem, strSvmParameters, 5);
//save trained model
svm_save_model(OutputModelFilename, pstrSvmModel);
//test trained model with training set -> train error
cout << "test model " << OutputModelFilename << " with the training set " << pInputSampleFileName << std::endl;
this->test(pInputSampleFileName, OutputModelFilename, dTrainError);
//clean up
svm_destroy_model(pstrSvmModel);
svm_destroy_param(&strSvmParameters);
free(strSvmProblem.y);
free(strSvmProblem.x);
dRetTrainError = dTrainError;
dRetCrossValError = dCrossValError;
}
bool Problem::CProblem::trainAndSaveModel(const FilePath& path, const Paramter& param) const
{
if (!m_hasData)
{
return false;
}
svm_model* model = svm_train(&m_problem, ¶m);
const FilePath parentFilePath = FileSystem::ParentPath(path);
if (!FileSystem::Exists(parentFilePath) && !FileSystem::CreateDirectories(parentFilePath))
{
return false;
}
const int32 result = svm_save_model(path.narrow().c_str(), model);
svm_free_and_destroy_model(&model);
return (result == 0);
}
int svmtrain(int argc, char **argv)
{
char input_file_name[1024];
char model_file_name[1024];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
LOGD("ERROR: %s\n",error_msg);
exit(1);
}
if(cross_validation)
{
do_cross_validation();
}
else
{
modelt = svm_train(&prob,¶m);
if(svm_save_model(model_file_name,modelt))
{
LOGD("can't save model to file %s\n", model_file_name);
exit(1);
}
svm_free_and_destroy_model(&modelt);
}
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
free(line);
return 0;
}
bool CmySvmArth::SaveTrainModel(const char* pPath)
{
if(pPath==NULL||model==NULL)
return false;
if(svm_save_model(pPath,model))
{
fprintf(stderr, "can't save model to file %s\n", pPath);
return false;
}
svm_free_and_destroy_model(&model);
if(x_space!=NULL)
{
free(x_space);
x_space = NULL;
}
if(line!=NULL)
{
free(line);
line = NULL;
}
return true;
}
void svmwrite (double *v, int *r, int *c,
int *rowindex,
int *colindex,
double *coefs,
double *rho,
int *compprob,
double *probA,
double *probB,
int *nclasses,
int *totnSV,
int *labels,
int *nSV,
int *sparsemodel,
int *svm_type,
int *kernel_type,
int *degree,
double *gamma,
double *coef0,
char **filename)
{
struct svm_model m;
int i;
char *fname = *filename;
/* set up model */
m.l = *totnSV;
m.nr_class = *nclasses;
m.sv_coef = (double **) malloc (m.nr_class * sizeof(double*));
for (i = 0; i < m.nr_class - 1; i++) {
m.sv_coef[i] = (double *) malloc (m.l * sizeof (double));
memcpy (m.sv_coef[i], coefs + i*m.l, m.l * sizeof (double));
}
if (*sparsemodel > 0)
m.SV = transsparse(v, *r, rowindex, colindex);
else
m.SV = sparsify(v, *r, *c);
m.rho = rho;
m.label = labels;
m.nSV = nSV;
if (*compprob) {
m.probA = probA;
m.probB = probB;
} else {
m.probA = NULL;
m.probB = NULL;
}
/* set up parameter */
m.param.svm_type = *svm_type;
m.param.kernel_type = *kernel_type;
m.param.degree = *degree;
m.param.gamma = *gamma;
m.param.coef0 = *coef0;
m.free_sv = 1;
/* write svm model */
svm_save_model(fname, &m);
for (i = 0; i < m.nr_class - 1; i++)
free(m.sv_coef[i]);
free(m.sv_coef);
for (i = 0; i < *r; i++)
free (m.SV[i]);
free (m.SV);
}
int main(int argc, char**argv)
{
if (argc < 3) {
printf("\nUSAGE: %s <ASCII model file> <binary model file>\n",argv[0]);
printf(" Convert an ASCII LibSVM model file into a binary LibSVM model file.\n\n");
printf("USAGE: %s - <binary model file> <ASCII model file>\n",argv[0]);
printf(" Convert a binary LibSVM model file into an ASCII LibSVM model file.\n\n");
return -1;
}
if (argv[1][0] == '-') { // bin -> ASCII
if (argc < 4) {
printf("USAGE: %s - <binary model file> <ASCII model file>\n",argv[0]);
printf(" Convert a binary LibSVM model file into an ASCII LibSVM model file.\n\n");
return -1;
}
printf("Loading binary model '%s'... ",argv[2]);
svm_model * m = svm_load_binary_model(argv[2]);
if (m==NULL) {
printf("\nERROR: failed loading model '%s'!\n",argv[2]);
return -2;
}
printf ("OK\n");
printf("Saving ASCII model '%s'... ",argv[3]);
int r = svm_save_model(argv[3],m);
if (!r) printf ("OK\n");
else {
printf("ERROR: failed saving ASCII model '%s' (code=%i)\n",argv[3],r);
return -3;
}
svm_destroy_model(m);
} else { // ASCII -> bin
printf("Loading ASCII model '%s'... ",argv[1]);
svm_model * m = svm_load_model(argv[1]);
if (m==NULL) {
printf("\nERROR: failed loading model '%s'!\n",argv[1]);
return -2;
}
printf ("OK\n");
printf("Saving binary model '%s'... ",argv[2]);
int r = svm_save_binary_model(argv[2],m);
if (!r) printf ("OK\n");
else {
printf("ERROR: failed saving binary model '%s' (code=%i)\n",argv[2],r);
return -3;
}
svm_destroy_model(m);
}
return 0;
}
void SVM::save_model(string model_name)
{
svm_save_model(model_name.c_str(), model);
}
Transformation SVMTrain::analyze(const DataSet* dataset) const {
G_INFO("Doing SVMTrain analysis...");
QStringList sdescs = selectDescriptors(dataset->layout(), StringType, _descriptorNames, _exclude, false);
if (!sdescs.isEmpty()) {
throw GaiaException("SVMTrain: if you want to use string descriptors for training your SVM, "
"you first need to enumerate them using the 'enumerate' transform on them. "
"String descriptors: ", sdescs.join(", "));
}
QStringList descs = selectDescriptors(dataset->layout(), UndefinedType, _descriptorNames, _exclude);
// sort descriptors in the order in which they are taken inside the libsvm dataset
sort(descs.begin(), descs.end(), DescCompare(dataset->layout()));
Region region = dataset->layout().descriptorLocation(descs);
QStringList classMapping = svm::createClassMapping(dataset, _className);
// first, convert the training data into an SVM problem structure
// NB: all checks about fixed-length and type of descriptors are done in this function
struct svm_problem prob = svm::dataSetToLibsvmProblem(dataset, _className, region, classMapping);
// also get dimension (this trick works because a vector in there is the number
// of dimensions + 1 for the sentinel, and we're not in a sparse representation)
int dimension = prob.x[1] - prob.x[0] - 1;
// default values
struct svm_parameter param;
//param.svm_type = C_SVC;
//param.kernel_type = RBF;
//param.degree = 3;
//param.gamma = 0; // 1/k
param.coef0 = 0;
param.nu = 0.5;
param.cache_size = 100;
//param.C = 1;
param.eps = 1e-3;
param.p = 0.1;
param.shrinking = 1;
//param.probability = 0;
param.nr_weight = 0;
param.weight_label = NULL;
param.weight = NULL;
// get parameters
QString svmType = _params.value("type", "C-SVC").toString().toLower();
param.svm_type = _svmTypeMap.value(svmType);
QString kernelType = _params.value("kernel", "RBF").toString().toLower();
param.kernel_type = _kernelTypeMap.value(kernelType);
param.degree = _params.value("degree", 3).toInt();
param.C = _params.value("c", 1).toDouble();
param.gamma = _params.value("gamma", 1.0/dimension).toDouble();
param.probability = _params.value("probability", false).toBool() ? 1 : 0;
const char* error_msg = svm_check_parameter(&prob, ¶m);
if (error_msg) {
throw GaiaException(error_msg);
}
// do it!
struct svm_model* model;
const bool crossValidation = false;
if (crossValidation) {
int nr_fold = 10;
int total_correct = 0;
double total_error = 0;
double sumv = 0, sumy = 0, sumvv = 0, sumyy = 0, sumvy = 0;
double* target = new double[prob.l];
svm_cross_validation(&prob, ¶m, nr_fold, target);
if (param.svm_type == EPSILON_SVR ||
param.svm_type == NU_SVR) {
for (int i=0; i<prob.l; i++) {
double y = prob.y[i];
double v = target[i];
total_error += (v-y)*(v-y);
sumv += v;
sumy += y;
sumvv += v*v;
sumyy += y*y;
sumvy += v*y;
}
G_INFO("Cross Validation Mean squared error =" << total_error/prob.l);
G_INFO("Cross Validation Squared correlation coefficient =" <<
((prob.l*sumvy - sumv*sumy) * (prob.l*sumvy - sumv*sumy)) /
((prob.l*sumvv - sumv*sumv) * (prob.l*sumyy - sumy*sumy))
);
}
else {
for (int i=0; i<prob.l; i++)
if (target[i] == prob.y[i])
++total_correct;
G_INFO("Cross Validation Accuracy =" << 100.0*total_correct/prob.l << "%");
}
}
else { // !crossValidation
model = svm_train(&prob, ¶m);
}
// save model to a temporary file (only method available from libsvm...),
// reload it and put it into a gaia2::Parameter
QTemporaryFile modelFile;
modelFile.open();
QString modelFilename = modelFile.fileName();
modelFile.close();
if (svm_save_model(modelFilename.toAscii().constData(), model) == -1) {
throw GaiaException("SVMTrain: error while saving SVM model to temp file");
}
modelFile.open();
QByteArray modelData = modelFile.readAll();
modelFile.close();
// if we asked for the model to be output specifically, also do it
if (_params.value("modelFilename", "").toString() != "") {
QString filename = _params.value("modelFilename").toString();
svm_save_model(filename.toAscii().constData(), model);
}
// destroy the model allocated by libsvm
svm_destroy_model(model);
Transformation result(dataset->layout());
result.analyzerName = "svmtrain";
result.analyzerParams = _params;
result.applierName = "svmpredict";
result.params.insert("modelData", modelData);
result.params.insert("className", _params.value("className"));
result.params.insert("descriptorNames", descs);
result.params.insert("classMapping", classMapping);
result.params.insert("probability", (param.probability == 1 && (param.svm_type == C_SVC ||
param.svm_type == NU_SVC)));
return result;
}
int main(int argc, char *argv[]) {
float labels[4] = {1.0, 1.0, -1.0, -1.0};
cv::Mat labelsMat(4, 1, CV_32FC1, labels);
float trainingData[4][2] = {{501, 10}, {255, 10},
{501, 255}, {10, 501}};
cv::Mat trainingDataMat(4, 2, CV_32FC1, trainingData);
svm_parameter param;
param.svm_type = C_SVC;
param.kernel_type = LINEAR;
param.degree = 3;
param.gamma = 0;
param.coef0 = 0;
param.nu = 0.5;
param.cache_size = 100;
param.C = 1;
param.eps = 1e-6;
param.p = 0.1;
param.shrinking = 1;
param.probability = 1;
param.nr_weight = 0;
param.weight_label = NULL;
param.weight = NULL;
svm_problem svm_prob_vector = libSVMWrapper(
trainingDataMat, labelsMat, param);
struct svm_model *model = new svm_model;
if (svm_check_parameter(&svm_prob_vector, ¶m)) {
std::cout << "ERROR" << std::endl;
} else {
model = svm_train(&svm_prob_vector, ¶m);
}
bool is_compute_probability = true;
std::string model_file_name = "svm";
bool save_model = true;
if (save_model) {
try {
svm_save_model(model_file_name.c_str(), model);
std::cout << "Model file Saved Successfully..." << std::endl;
} catch(std::exception& e) {
std::cout << e.what() << std::endl;
}
}
bool is_probability_model = svm_check_probability_model(model);
int svm_type = svm_get_svm_type(model);
int nr_class = svm_get_nr_class(model); // number of classes
double *prob_estimates = new double[nr_class];
cv::Vec3b green(0, 255, 0);
cv::Vec3b blue(255, 0, 0);
int width = 512, height = 512;
cv::Mat image = cv::Mat::zeros(height, width, CV_8UC3);
for (int i = 0; i < image.rows; ++i) {
for (int j = 0; j < image.cols; ++j) {
cv::Mat sampleMat = (cv::Mat_<float>(1, 2) << j, i);
int dims = sampleMat.cols;
svm_node* test_pt = new svm_node[dims];
for (int k = 0; k < dims; k++) {
test_pt[k].index = k + 1;
test_pt[k].value = static_cast<double>(sampleMat.at<float>(0, k));
}
test_pt[dims].index = -1;
float response = 0.0f;
if (is_probability_model && is_compute_probability) {
response = svm_predict_probability(model, test_pt, prob_estimates);
} else {
response = svm_predict(model, test_pt);
}
/*
std::cout << "Predict: " << prob << std::endl;
for (int y = 0; y < nr_class; y++) {
std::cout << prob_estimates[y] << " ";
}std::cout << std::endl;
*/
if (prob_estimates[0] > 0.5 || response == 1) {
image.at<cv::Vec3b>(i, j) = green;
} else if (prob_estimates[1] >= 0.5 || response == -1) {
image.at<cv::Vec3b>(i, j) = blue;
}
}
}
cv::imshow("image", image);
cv::waitKey(0);
return 0;
}
int
svmTrain (double gamma, std::vector<std::vector<svm_node> > data, std::vector<int> labels)
{
struct svm_model* model;
struct svm_problem prob;
struct svm_node* x_space;
struct svm_parameter param;
// Make sure we have data
if (data.empty ())
{
CORE_ERROR ("No training data\n");
return (-1);
}
prob.l = data.size ();
prob.y = Malloc (double, prob.l);
prob.x = Malloc (struct svm_node* , prob.l);
x_space = Malloc (struct svm_node, data.size () * data[0].size ());
int j = 0;
for (int i = 0; i < prob.l; ++i)
{
prob.y[i] = *(labels.begin () + i);
prob.x[i] = &x_space[j];
for (std::vector<svm_node>::iterator it = data[i].begin (); it != data[i].end (); ++it)
{
x_space[j].index = it->index;
x_space[j].value = it->value;
++j;
}
}
param.svm_type = C_SVC;
param.kernel_type = RBF;
param.gamma = gamma;
param.degree = 3;
param.coef0 = 0;
param.nu = 0.5;
param.cache_size = 100;
param.C = 1;
param.eps = 1e-3;
param.p = 0.1;
param.shrinking = 1;
param.probability = 0;
param.nr_weight = 0;
param.weight_label = NULL;
param.weight = NULL;
// Check whether the parameters are within a feasible range of the problem
const char* error_msg = svm_check_parameter (&prob, ¶m);
if (error_msg)
{
CORE_ERROR ("Error checking SVM parameters: %s\n", error_msg);
return (-1);
}
model = svm_train (&prob, ¶m);
if (svm_save_model ("model.txt", model))
{
CORE_ERROR ("Save SVM model failed\n");
return (-1);
}
svm_free_and_destroy_model (&model);
svm_destroy_param (¶m);
free (prob.y);
free (prob.x);
free (x_space);
return (0);
}
void LibsvmSvm::save(std::string file) {
svm_save_model(file.c_str(), _model);
Util::zip(file);
}
int Solver_Core::Solve(int num_basis, double cvm_eps)
{
timeusedForModelSaving = 0;
modelsWritten = 0;
algorithmStartTime = pnow();
nextSaveTime = param->saveFactor * (modelsWritten+1) + floor (pow (param->saveExponential, 1));
saveExponential = param->saveExponential;
this->maxNumBasis = num_basis;
// The convergence of CVM does not require the exact MEB on the core-set.
// With the recent advance of core-set approximation, the (1+epsilon/2)-MEB approximation on the core-set
// can guarantee the quality and the convergence of the (1+epsilon)-MEB approximation for the whole data set.
// See [Kumar, Mitchell, and Yildirim, 2003]
//
// iterate on epsilons
double epsilonFactor = EPS_SCALING;
for(double currentEpsilon = INITIAL_EPS; currentEpsilon/epsilonFactor > cvm_eps; currentEpsilon *= epsilonFactor)
{
// check epsilon
currentEpsilon = (currentEpsilon < cvm_eps ? cvm_eps : currentEpsilon);
// solve problem with current epsilon (warm start from the previous solution)
double maxDistance2 = 0.0;
int maxDistance2Idx = 0;
double factor = 1.0 + currentEpsilon;
factor *= factor;
// The convergence of CVM does not require the exact most violating point.
// A violating point is enough for the convergence of CVM.
// Probabilistic speedup lead to a good tradeoff between convergence and complexity at each iteration
//
while (maxDistance2Idx != -1)
{
// get a probabilistic sample
maxDistance2 = r2 * factor;
maxDistance2Idx = -1;
for(int sampleIter = 0; (sampleIter < 7) && (maxDistance2Idx == -1); sampleIter++)
maxDistance2 = _maxDistFromSampling(maxDistance2, maxDistance2Idx);
// check maximal distance
if (maxDistance2Idx != -1)
{
_UpdateCoreSet(maxDistance2Idx);
#ifndef RELEASE_VER
printf("#%d eps: %g |c|: %.10f R: %.10f |c-x|: %.10f r: %.10f\n",coreNum, currentEpsilon, coreNorm2, r2, maxDistance2, sqrt(maxDistance2/r2)-1.0);
#endif
solver->Solve(coreIdx,coreNum,tempD);
outAlpha = solver->getAlpha();
ComputeRadius2();
// if (coreNum%20 < 1) info(".");
}
double currentTime = pnow ();
if ((param->saveExponential >= 0.0) && (currentTime - algorithmStartTime >= nextSaveTime + timeusedForModelSaving )) {
modelsWritten++;
printf( "&%d..", modelsWritten);
nextSaveTime = param->saveFactor * (modelsWritten+1) + pow (param->saveExponential, modelsWritten + 1);
// std::cout << "Algorithm time is now " << currentTime - algorithmStartTime - timeusedForModelSaving << "\n";
// std::cout << " Saving model now at " << currentTime - startTime << "\n";
double before = pnow();
svm_model *model = Malloc(svm_model,1);
model->param = *param;
model->free_sv = 0; // XXX
model->nr_class = param->nr_classes;
model->label = Malloc(int,model->nr_class);
for(int i=0;i<model->nr_class;i++)
model->label[i] = param->label[i];
model->rho = Malloc(double,1);
double *alpha = Malloc(double,param->prob->l);
double THRESHOLD = 1e-5/coreNum;
for(int i=0;i<1;i++)
model->rho[i] = -ComputeSolution(alpha, THRESHOLD); //overwrite alphas!
model->nSV = Malloc(int, model->nr_class);
// assume all the nonzeros are false at the beginning,
// we have binary classification, they will not change in the
// outer loop in svm.cpp
int total_sv = 0;
for(int i=0;i<model->nr_class;i++)
{
int nSV = 0;
for(int j=0;j<param->count[i];j++) {
if(fabs(alpha[param->start[i]+j])) {
++nSV;
++total_sv;
}
}
model->nSV[i] = nSV;
}
model->l = total_sv; //anzahl der sv im model
// ASSUME PERMUTATION IS IDENTITY
svm_node **x = Malloc(svm_node *,param->prob->l);
int i;
for(int i=0;i<param->prob->l;i++)
x[i] = param->prob->x[i]; // x[i] = param->prob->x[perm[i]];
model->SV = Malloc(svm_node *,total_sv);
int p = 0;
for(int i=0;i<param->prob->l;i++) {
if(fabs(alpha[i]) > 0.0) {
model->SV[p++] = x[i];
}
}
model->sv_coef = Malloc(double *,model->nr_class-1);
for(int i=0;i<model->nr_class-1;i++) {
model->sv_coef[i] = Malloc(double,total_sv);
}
// printf ("%d SV class 0, %d SV class 1\n", model->nSV[0], model->nSV[1]);
int q = 0;
for(int i=0;i<model->nr_class;i++)
{
for(int j=0;j < param->count[i];j++) {
if(fabs(alpha[param->start[i]+j]) > 0.0) {
model->sv_coef[0][q++] = alpha[param->start[i]+j];// * param->prob->y[param->start[i]+j];
// printf ("%d -- %f\n", q, alpha[param->start[i]+j]);
}
}
}
p = 0;
char tmp[1000];
if (strlen(param->modelPath) < 2) {
sprintf(tmp,"%s_%d",param->modelFile, (int)floor( before - algorithmStartTime - timeusedForModelSaving) );
} else {
sprintf(tmp,"%s/%d.cvm.model",param->modelPath, (int)floor( before - algorithmStartTime - timeusedForModelSaving) ) ;
}
model->probB = NULL;
model->probA = NULL;
svm_save_model(tmp, model);
// just to keep compability
FILE* fModel = fopen(tmp, "a+t"); // append mode
fprintf(fModel, "CPU Time = %f second\n", before - algorithmStartTime - timeusedForModelSaving);
fclose(fModel);
// svm_destroy_model(model);
//free(x);
double after = pnow();
timeusedForModelSaving += after - before;
}
// check if we are over the walltime
if (pnow() - param->startTime > param->wallTime + timeusedForModelSaving) {
printf ("Hit the walltime, exiting now..\n");
return coreNum;
}
if (IsExitOnMaxIter())
{
return coreNum;
break;
}
}
}
//train the svm using the parameters defined inside this method
void ML2::trainData()
{
//file to store the svm model structure
string model_file_name1 = flowstatistics_train_name+"_model.csv";
const char *model_file_name = model_file_name1.c_str();
//file to read the data from
char input_file_name[1024] = "velocityArray.csv";
//char input_file_name2[1024] = "data/flowstatistics_train_mu.csv";
const char *error_msg;
//parameters of the svm
/*
"-s svm_type : set type of SVM (default 0)\n"
" 0 -- C-SVC (multi-class classification)\n"
" 1 -- nu-SVC (multi-class classification)\n"
" 2 -- one-class SVM\n"
" 3 -- epsilon-SVR (regression)\n"
" 4 -- nu-SVR (regression)\n"
"-t kernel_type : set type of kernel function (default 2)\n"
" 0 -- linear: u'*v\n"
" 1 -- polynomial: (gamma*u'*v + coef0)^degree\n"
" 2 -- radial basis function: exp(-gamma*|u-v|^2)\n"
" 3 -- sigmoid: tanh(gamma*u'*v + coef0)\n"
" 4 -- precomputed kernel (kernel values in training_set_file)\n"
"-d degree : set degree in kernel function (default 3)\n"
"-g gamma : set gamma in kernel function (default 1/num_features)\n"
"-r coef0 : set coef0 in kernel function (default 0)\n"
"-c cost : set the parameter C of C-SVC, epsilon-SVR, and nu-SVR (default 1)\n"
"-n nu : set the parameter nu of nu-SVC, one-class SVM, and nu-SVR (default 0.5)\n"
"-p epsilon : set the epsilon in loss function of epsilon-SVR (default 0.1)\n"
"-m cachesize : set cache memory size in MB (default 100)\n"
"-e epsilon : set tolerance of termination criterion (default 0.001)\n"
"-h shrinking : whether to use the shrinking heuristics, 0 or 1 (default 1)\n"
"-b probability_estimates : whether to train a SVC or SVR model for probability estimates, 0 or 1 (default 0)\n"
"-wi weight : set the parameter C of class i to weight*C, for C-SVC (default 1)\n"
"-v n: n-fold cross validation mode\n"
"-q : quiet mode (no outputs)\n"
*/
//set the parameters to be used for svm
param.svm_type = 4;
param.kernel_type = 1;//RBF;
param.degree = 2;
param.gamma = 0.125; // 1/num_features
param.coef0 = 0;
param.nu = 0.4;
param.cache_size = 100;
param.C = 0.125;
param.eps = 1e-3;
param.p = 0.1;
param.shrinking = 1;
param.probability = 0;
param.nr_weight = 0;
param.weight_label = NULL;
param.weight = NULL;
//param.v = 10;
nr_fold =10;
//read from the data file
read_problem_data( input_file_name, flowcol);
//checking the parameters, if they are set correctly
error_msg = svm_check_parameter(&prob,¶m);
if(error_msg)
{
cout<<"ERROR: "<<error_msg<<endl<<flush;
exit(1);
}
//do_cross_validation();
// first do grid search do find optimal parameters
//paramSelection();
// then do training with optimal parameters
//param.gamma = best_gamma;
//param.C = best_C;
cout<< "start training\n"<<endl<<flush;
model = svm_train(&prob,¶m);
cout<< "end training\n"<<endl<<flush;
// then do cross fold validation
cout<< "start with cross validation" <<endl<<flush;
do_cross_validation();
cout<< "end cross validation" <<endl<<flush;
//save model
if(svm_save_model(model_file_name,model))
{
cout<< "can't save model to file "<< model_file_name <<endl<<flush;
exit(1);
}
//free all the pointers used, except the model which is required for prediction
// svm_destroy_param(¶m);
// free(prob.y);
// free(prob.x);
// free(x_space);
// free(line);
return;
}
int main(int argc, char **argv)
{
#ifdef WIN32
// Send all reports to STDOUT
_CrtSetReportMode( _CRT_WARN, _CRTDBG_MODE_FILE );
_CrtSetReportFile( _CRT_WARN, _CRTDBG_FILE_STDOUT );
_CrtSetReportMode( _CRT_ERROR, _CRTDBG_MODE_FILE );
_CrtSetReportFile( _CRT_ERROR, _CRTDBG_FILE_STDOUT );
_CrtSetReportMode( _CRT_ASSERT, _CRTDBG_MODE_FILE );
_CrtSetReportFile( _CRT_ASSERT, _CRTDBG_FILE_STDOUT );
// enable the options
SET_CRT_DEBUG_FIELD( _CRTDBG_DELAY_FREE_MEM_DF );
SET_CRT_DEBUG_FIELD( _CRTDBG_LEAK_CHECK_DF );
#endif
printf("int %d, short int %d, char %d, double %d, float %d, node %d\n",sizeof(int),sizeof(short int), sizeof(char), sizeof(double), sizeof(float), sizeof(svm_node));
char input_file_name[FILENAME_LEN];
char model_file_name[FILENAME_LEN];
const char *error_msg;
parse_command_line(argc, argv, input_file_name, model_file_name);
read_problem(input_file_name);
param.modelFile = model_file_name;
printf ("Finish reading input files!\n");
error_msg = svm_check_parameter(&prob,¶m);
#ifdef WIN32
assert(_CrtCheckMemory());
#endif
if(error_msg)
{
fprintf(stderr,"Error: %s\n",error_msg);
exit(1);
}
double duration;
double start = getRunTime();
if(cross_validation)
{
do_cross_validation();
}
else
{
printf("kernel: %d\n",param.kernel_type);
model = svm_train(&prob,¶m);
double finish = getRunTime();
duration = (double)(finish - start);
#ifdef WIN32
assert(_CrtCheckMemory());
#endif
svm_save_model(model_file_name,model);
svm_destroy_model(model);
}
printf("CPU Time = %f second\n", duration);
FILE* fModel = fopen(model_file_name, "a+t"); // append mode
fprintf(fModel, "CPU Time = %f second\n", duration);
fclose(fModel);
svm_destroy_param(¶m);
free(prob.y);
free(prob.x);
free(x_space);
#ifdef WIN32
assert(_CrtCheckMemory());
#endif
return 0;
}
/* @return -> cross validation error
* -1 -> could not open statistic_filename
*
*/
int SVM::crossValidationParameters(char* pSampleFilename, char* pFixedTestFilename, char* pStatisticsFilename)
{
struct svm_parameter strSvmParameters;
struct svm_problem strSvmProblem;
struct svm_model* pstrSvmModel;
const char *error_msg;
double dCrossValError = -1;
double dFixTestSetError = -1;
double dTrainError = -1;
ofstream outputStatisticFile;
char *pTempModelFilename = "tempSvmModel.tmp";
double dGamma = -1;
double dC = -1;
char tempText[1000];
//open statistic file
outputStatisticFile.open(pStatisticsFilename, std::ios::trunc);
if(!outputStatisticFile.is_open())
{
cout << "can not open statistic file " << pStatisticsFilename << std::endl;
return -1;
}
//initial parameters
this->setParameters(strSvmParameters);
this->read_problem(pSampleFilename, strSvmProblem, strSvmParameters);
error_msg = svm_check_parameter(&strSvmProblem, &strSvmParameters);
outputStatisticFile << "\% " << pSampleFilename << std::endl;
outputStatisticFile << "\% Gamma \t C \t TrainError \t CrossValError \t FixedTestError" << std::endl;
cout << "\% Gamma \t C \t TrainError \t CrossValError \t FixedTestError" << std::endl;
for(int iExponentGamma = -25; iExponentGamma <= -2; ++iExponentGamma)
{
dGamma = pow(2.0, iExponentGamma);
strSvmParameters.gamma = dGamma;
for(int iExponentC = -1; iExponentC <= 15; ++iExponentC)
{
dC = pow(2.0, iExponentC);
strSvmParameters.C = dC;
//train
pstrSvmModel = svm_train(&strSvmProblem, &strSvmParameters);
svm_save_model(pTempModelFilename, pstrSvmModel);
svm_destroy_model(pstrSvmModel);
//test with train-set
this->test(pSampleFilename, pTempModelFilename, dTrainError);
//crossval
dCrossValError = this->crossValidationSamples(strSvmProblem, strSvmParameters, 5);
//test with fixed test-set
this->test(pFixedTestFilename, pTempModelFilename, dFixTestSetError);
remove(pTempModelFilename);
sprintf(tempText, "%.15lf\t%10lf\t%10lf\t%10lf\t%10lf", dGamma, dC, dTrainError, dCrossValError, dFixTestSetError);
outputStatisticFile << tempText << std::endl;
cout << "\n#######################################################################" << std::endl;
cout << tempText << std::endl;
cout << "#######################################################################\n" << std::endl;
}
}
//clean up
outputStatisticFile.close();
svm_destroy_param(&strSvmParameters);
free(strSvmProblem.y);
free(strSvmProblem.x);
//free(pTempModelFilename);
return 0;
}
double im_svm_train(IMbdd& bdd,
const std::vector<std::string>& trainingPeople,
MatrixC& trainMC,
const std::vector<std::string>& testingPeople,
MatrixC& testMC){
std::string path2bdd(bdd.getFolder());
int k = bdd.getK();
// Computing Bag Of Words for each people
std::cout << "Computing BOWs..." << std::endl;
std::map<std::string, struct svm_problem> peopleBOW;
std::map<std::string, int> activitiesLabel;
std::cout << "Computing the SVM problem (ie. BOW) of all the people..." << std::endl;
im_compute_bdd_bow(bdd,peopleBOW); // all the BOW are saved in peopleBOW
// Normalization: do not forget to change the normalization before this step
//bdd.changeNormalizationSettings(normalization,
// "means.txt",
//"stand_devia.txt");
for(std::vector<std::string>::const_iterator trainingPerson = trainingPeople.begin();
trainingPerson != trainingPeople.end();
++ trainingPerson){
for(int i=0 ; i<peopleBOW[*trainingPerson].l ; i++)
std::cout << peopleBOW[*trainingPerson].y[i] << std::endl;
}
std::cout << "Normalizing the BOW..." << std::endl;
bdd.changeNormalizationSettings("simple",
"means.txt",
"stand_devia.txt");
im_normalize_bdd_bow(bdd,trainingPeople,peopleBOW);
std::cout << "Not exporting the problem..." << std::endl;
// Export ?? OR NOT export ???????? THAT IS THE QUESTION
// Exporting problem
//exportProblem(svmTrainProblem, path2bdd + "/concatenate.bow.train");
//if(testingPeople != {""})
//exportProblem(svmTestProblem, path2bdd + "/concatenate.bow.test");
// * SVM *
// (One Versus the Rest (OVR))
// SVM parameters
struct svm_parameter svmParameter;
get_svm_parameter(k,svmParameter);
std::cout << "Searching the best C and the best Gamma..." << std::endl;
// We have to do a leave one out with the training data
// For the moment we use the same training centers to find Gamma and C
int minC = -5, maxC = 5;
int minG = -10, maxG = 3;
double crossValidationAccuracy =
im_training_leave_one_out(bdd,
trainingPeople, peopleBOW,
minC, maxC, // ln(min/max C)
minG, maxG, // ln(min/max G)
svmParameter);
// N.B: when we export the model to the robot, peopleBOW index = trainingPeople
// Generating the SVM model
std::cout << "Generating the SVM model..." << std::endl;
struct svm_problem trainingProblem;
trainingProblem.l = 0;
trainingProblem.x = NULL;
trainingProblem.y = NULL;
for(std::vector<std::string>::const_iterator trainingPerson = trainingPeople.begin();
trainingPerson != trainingPeople.end();
++ trainingPerson){
for(int i=0 ; i < peopleBOW.at(*trainingPerson).l ; i++){
struct svm_node* bow = peopleBOW.at(*trainingPerson).x[i];
addBOW(bow, peopleBOW.at(*trainingPerson).y[i],trainingProblem);
}
}
struct svm_model** svmModels = svm_train_ovr(&trainingProblem,&svmParameter);
// Exporting models
std::cout << "Saving the SVM model..." << std::endl;
std::vector <std::string> modelFiles;
int nrActivities = bdd.getActivities().size();
for(int i=0 ; i< nrActivities ; i++){
std::string fileToSaveModel = path2bdd;
std::stringstream ss;
ss << i;
fileToSaveModel = fileToSaveModel + "/svm_ovr_" + ss.str() + ".model";
svm_save_model(fileToSaveModel.c_str(),svmModels[i]);
modelFiles.push_back(fileToSaveModel);
}
bdd.changeSVMSettings(nrActivities,
modelFiles);
// Calculate the confusion matrix and the probability estimation
std::cout << "Filling the training confusion matrix..." << std::endl;
im_fill_confusion_matrix(bdd,trainingProblem,svmModels, trainMC);
destroy_svm_problem(trainingProblem);
if(testingPeople.size() > 0){
std::cerr << "Entering in Hell..." << std::endl;
struct svm_problem testingProblem;
trainingProblem.l = 0;
trainingProblem.x = NULL;
trainingProblem.y = NULL;
for(std::vector<std::string>::const_iterator testingPerson = testingPeople.begin();
testingPerson != testingPeople.end();
++ testingPerson){
for(int i=0 ; i < peopleBOW.at(*testingPerson).l ; i++){
std::cerr << "HELlllll" << std::endl;
struct svm_node* bow = peopleBOW.at(*testingPerson).x[i];
std::cerr << peopleBOW.at(*testingPerson).y[i] << std::endl;
std::cerr << "YOOOO man" << std::endl;
addBOW(bow, peopleBOW.at(*testingPerson).y[i],testingProblem);
std::cerr << "Why ?!!!" << std::endl;
}
}
std::cout << "Filling the testing confusion matrix..." << std::endl;
im_fill_confusion_matrix(bdd,testingProblem,svmModels, testMC);
destroy_svm_problem(testingProblem);
}
std::cout << "Releasing peopleBOW" << std::endl;
// Releasing peopleBOW
std::vector<std::string> people = bdd.getPeople();
for(std::vector<std::string>::iterator person = people.begin();
person != people.end();
++ person){
destroy_svm_problem(peopleBOW[*person]);
}
// Releasing OVR models
for(int i=0;i<nrActivities;i++){
svm_free_and_destroy_model(&svmModels[i]);}
delete [] svmModels;
svmModels = NULL;
return crossValidationAccuracy;
}