/usr/include/OpenMS/ANALYSIS/SVM/SVMWrapper.h is in libopenms-dev 1.11.1-5.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 | // --------------------------------------------------------------------------
// OpenMS -- Open-Source Mass Spectrometry
// --------------------------------------------------------------------------
// Copyright The OpenMS Team -- Eberhard Karls University Tuebingen,
// ETH Zurich, and Freie Universitaet Berlin 2002-2013.
//
// This software is released under a three-clause BSD license:
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
// * Neither the name of any author or any participating institution
// may be used to endorse or promote products derived from this software
// without specific prior written permission.
// For a full list of authors, refer to the file AUTHORS.
// --------------------------------------------------------------------------
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL ANY OF THE AUTHORS OR THE CONTRIBUTING
// INSTITUTIONS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// --------------------------------------------------------------------------
// $Maintainer: Sandro Andreotti $
// $Authors: Nico Pfeifer, Chris Bielow $
// --------------------------------------------------------------------------
#ifndef OPENMS_ANALYSIS_SVM_SVMWRAPPER_H
#define OPENMS_ANALYSIS_SVM_SVMWRAPPER_H
#include <svm.h>
#include <OpenMS/CONCEPT/Types.h>
#include <OpenMS/CONCEPT/ProgressLogger.h>
#include <OpenMS/DATASTRUCTURES/String.h>
#include <OpenMS/FORMAT/TextFile.h>
#include <OpenMS/SYSTEM/File.h>
#include <string>
#include <vector>
#include <map>
#include <cmath>
#include <iostream>
#include <fstream>
namespace OpenMS
{
/// Data structure used in SVMWrapper
struct SVMData
{
std::vector<std::vector<std::pair<Int, DoubleReal> > > sequences;
std::vector<DoubleReal> labels;
SVMData() :
sequences(std::vector<std::vector<std::pair<Int, DoubleReal> > >()),
labels(std::vector<DoubleReal>())
{
}
SVMData(std::vector<std::vector<std::pair<Int, DoubleReal> > > & seqs, std::vector<DoubleReal> & lbls) :
sequences(seqs),
labels(lbls)
{
}
bool operator==(const SVMData & rhs) const
{
return sequences == rhs.sequences
&& labels == rhs.labels;
}
bool store(const String & filename) const
{
std::ofstream output_file(filename.c_str());
// checking if file is writable
if (!File::writable(filename) || sequences.size() != labels.size())
{
return false;
}
// writing feature vectors
for (Size i = 0; i < sequences.size(); i++)
{
output_file << labels[i] << " ";
for (Size j = 0; j < sequences[i].size(); ++j)
{
output_file << sequences[i][j].second << ":" << sequences[i][j].first << " ";
}
output_file << std::endl;
}
output_file.flush();
output_file.close();
std::cout.flush();
return true;
}
bool load(const String & filename)
{
Size counter = 0;
std::vector<String> parts;
std::vector<String> temp_parts;
if (!File::exists(filename))
{
return false;
}
if (!File::readable(filename))
{
return false;
}
if (File::empty(filename))
{
return false;
}
TextFile text_file(filename.c_str(), true);
TextFile::iterator it;
it = text_file.begin();
sequences.resize(text_file.size(), std::vector<std::pair<Int, DoubleReal> >());
labels.resize(text_file.size(), 0.);
while (counter < text_file.size() && it != text_file.end())
{
it->split(' ', parts);
labels[counter] = parts[0].trim().toFloat();
sequences[counter].resize(parts.size(), std::pair<Int, DoubleReal>());
for (Size j = 1; j < parts.size(); ++j)
{
parts[j].split(':', temp_parts);
if (temp_parts.size() < 2)
{
return false;
}
sequences[counter][j - 1].second = temp_parts[0].trim().toFloat();
sequences[counter][j - 1].first = temp_parts[1].trim().toInt();
}
++counter;
++it;
}
return true;
}
};
/**
@brief Serves as a wrapper for the libsvm
This class can be used for svm predictions. You can either perform classification or regression and
choose certain kernel fuctions and additional parameters. Furthermore the models can be saved and
loaded and we support also a new kernel function that was specially designed for learning with
small sequences of different lengths.
*/
class OPENMS_DLLAPI SVMWrapper :
public ProgressLogger
{
public:
/**
@brief Parameters for the svm to be set from outside
This type is used to specify the kind of parameter that
is to be set or retrieved by the set/getParameter methods.
*/
enum SVM_parameter_type
{
SVM_TYPE, ///< the svm type cab be NU_SVR or EPSILON_SVR
KERNEL_TYPE, ///< the kernel type
DEGREE, ///< the degree for the polynomial- kernel
C, ///< the C parameter of the svm
NU, ///< the nu parameter for nu-SVR
P, ///< the epsilon parameter for epsilon-SVR
GAMMA, ///< the gamma parameter of the POLY, RBF and SIGMOID kernel
PROBABILITY, ///<
SIGMA, ///<
BORDER_LENGTH ///<
};
/// Kernel type
enum SVM_kernel_type
{
OLIGO = 19,
OLIGO_COMBINED
};
/// standard constructor
SVMWrapper();
/// destructor
virtual ~SVMWrapper();
/**
@brief You can set the parameters of the svm:
KERNEL_TYPE: can be LINEAR for the linear kernel
RBF for the rbf kernel
POLY for the polynomial kernel
SIGMOID for the sigmoid kernel
DEGREE: the degree for the polynomial- kernel and the
locality- improved kernel
C: the C parameter of the svm
*/
void setParameter(SVM_parameter_type type, Int value);
/**
@brief sets the double parameters of the svm
*/
void setParameter(SVM_parameter_type type, DoubleReal value);
/**
@brief trains the svm
The svm is trained with the data stored in the 'svm_problem' structure.
*/
Int train(struct svm_problem * problem);
/**
@brief trains the svm
The svm is trained with the data stored in the 'SVMData' structure.
*/
Int train(SVMData & problem);
/**
@brief saves the svm model
The model of the trained svm is saved into 'modelFilename'. Throws an exception if
the model cannot be saved.
@exception Exception::UnableToCreateFile
*/
void saveModel(std::string modelFilename) const;
/**
@brief loads the model
The svm- model is loaded. After this, the svm is ready for
prediction.
*/
void loadModel(std::string modelFilename);
/**
@brief predicts the labels using the trained model
The prediction process is started and the results are stored in 'predicted_labels'.
*/
void predict(struct svm_problem * problem, std::vector<DoubleReal> & predicted_labels);
/**
@brief predicts the labels using the trained model
The prediction process is started and the results are stored in 'predicted_labels'.
*/
void predict(const SVMData & problem, std::vector<DoubleReal> & results);
/**
@brief You can get the actual int- parameters of the svm
KERNEL_TYPE: can be LINEAR for the linear kernel
RBF for the rbf kernel
POLY for the polynomial kernel
SIGMOID for the sigmoid kernel
DEGREE: the degree for the polynomial- kernel and the
locality- improved kernel
SVM_TYPE: the SVm type of the svm: can be NU_SVR or EPSILON_SVR
*/
Int getIntParameter(SVM_parameter_type type);
/**
@brief You can get the actual double- parameters of the svm
C: the C parameter of the svm
P: the P parameter of the svm (sets the epsilon in
epsilon-svr)
NU: the nu parameter in nu-SVR
GAMMA: for POLY, RBF and SIGMOID
*/
DoubleReal getDoubleParameter(SVM_parameter_type type);
/**
@brief You can create 'number' equally sized random partitions
This function creates 'number' equally sized random partitions and stores them in 'partitions'.
*/
static void createRandomPartitions(svm_problem * problem, Size number, std::vector<svm_problem *> & partitions);
/**
@brief You can create 'number' equally sized random partitions
This function creates 'number' equally sized random partitions and stores them in 'partitions'.
*/
static void createRandomPartitions(const SVMData & problem,
Size number,
std::vector<SVMData> & problems);
/**
@brief You can merge partitions excuding the partition with index 'except'
*/
static svm_problem * mergePartitions(const std::vector<svm_problem *> & problems, Size except);
/**
@brief You can merge partitions excuding the partition with index 'except'
*/
static void mergePartitions(const std::vector<SVMData> & problems,
Size except,
SVMData & merged_problem);
/**
@brief predicts the labels using the trained model
The prediction process is started and the results are stored in 'predicted_rts'.
*/
void predict(const std::vector<svm_node *> & vectors, std::vector<DoubleReal> & predicted_rts);
/**
@brief Stores the stored labels of the encoded SVM data at 'labels'
*/
static void getLabels(svm_problem * problem, std::vector<DoubleReal> & labels);
/**
@brief Performs a CV for the data given by 'problem'
*/
DoubleReal performCrossValidation(svm_problem * problem_ul,
const SVMData & problem_l,
const bool is_labeled,
const std::map<SVM_parameter_type, DoubleReal> & start_values_map,
const std::map<SVM_parameter_type, DoubleReal> & step_sizes_map,
const std::map<SVM_parameter_type, DoubleReal> & end_values_map,
Size number_of_partitions,
Size number_of_runs,
std::map<SVM_parameter_type, DoubleReal> & best_parameters,
bool additive_step_sizes = true,
bool output = false,
String performances_file_name = "performances.txt",
bool mcc_as_performance_measure = false);
/**
@brief Returns the probability parameter sigma of the fitted laplace model.
The libsvm is used to fit a laplace model to the prediction values by performing
an internal cv using the training set if setParameter(PROBABILITY, 1) was invoked
before using train. Look for your libsvm documentation for more details.
The model parameter sigma is returned by this method. If no model was fitted during
training zero is returned.
*/
DoubleReal getSVRProbability();
/**
@brief returns the value of the oligo kernel for sequences 'x' and 'y'
This function computes the kernel value of the oligo kernel,
which was introduced by Meinicke et al. in 2004. 'x' and
'y' are encoded by encodeOligo and 'gauss_table' has to be
constructed by calculateGaussTable.
'max_distance' can be used to speed up the computation
even further by restricting the maximum distance between a k_mer at
position i in sequence 'x' and a k_mer at position j
in sequence 'y'. If i - j > 'max_distance' the value is not
added to the kernel value. This approximation is switched
off by default (max_distance < 0).
*/
static DoubleReal kernelOligo(const std::vector<std::pair<int, double> > & x,
const std::vector<std::pair<int, double> > & y,
const std::vector<double> & gauss_table,
int max_distance = -1);
/**
@brief calculates the oligo kernel value for the encoded sequences 'x' and 'y'
This kernel function calculates the oligo kernel value [Meinicke 04] for
the sequences 'x' and 'y' that had been encoded by the encodeOligoBorder... function
of the LibSVMEncoder class.
*/
static DoubleReal kernelOligo(const svm_node * x, const svm_node * y, const std::vector<DoubleReal> & gauss_table, DoubleReal sigma_square = 0, Size max_distance = 50);
/**
@brief calculates the significance borders of the error model and stores them in 'sigmas'
*/
void getSignificanceBorders(svm_problem * data, std::pair<DoubleReal, DoubleReal> & borders, DoubleReal confidence = 0.95, Size number_of_runs = 5, Size number_of_partitions = 5, DoubleReal step_size = 0.01, Size max_iterations = 1000000);
/**
@brief calculates the significance borders of the error model and stores them in 'sigmas'
*/
void getSignificanceBorders(const SVMData & data,
std::pair<DoubleReal, DoubleReal> & sigmas,
DoubleReal confidence = 0.95,
Size number_of_runs = 5,
Size number_of_partitions = 5,
DoubleReal step_size = 0.01,
Size max_iterations = 1000000);
/**
@brief calculates a p-value for a given data point using the model parameters
Uses the model parameters to calculate the p-value for 'point' which has the data
entries: measured, predicted retention time.
*/
DoubleReal getPValue(DoubleReal sigma1, DoubleReal sigma2, std::pair<DoubleReal, DoubleReal> point);
/**
@brief stores the prediction values for the encoded data in 'decision_values'
This function can be used to get the prediction values of the data if a model
is already trained by the train() method. For regression the result is the same
as for the method predict. For classification this function returns the distance from
the separating hyperplane. For multiclass classification the decision_values vector
will be empty.
*/
void getDecisionValues(svm_problem * data, std::vector<DoubleReal> & decision_values);
/**
@brief Scales the data such that every coloumn is scaled to [-1, 1].
Scales the x[][].value values of the svm_problem* structure. If the second
parameter is omitted, the data is scaled to [-1, 1]. Otherwise the data is scaled to [0, max_scale_value]
*/
void scaleData(svm_problem * data, Int max_scale_value = -1);
static void calculateGaussTable(Size border_length, DoubleReal sigma, std::vector<DoubleReal> & gauss_table);
/**
@brief computes the kernel matrix using the actual svm parameters and the given data
This function can be used to compute a kernel matrix. 'problem1' and 'problem2'
are used together wit the oligo kernel function (could be extended if you
want to use your own kernel functions).
*/
svm_problem * computeKernelMatrix(svm_problem * problem1, svm_problem * problem2);
/**
@brief computes the kernel matrix using the actual svm parameters and the given data
This function can be used to compute a kernel matrix. 'problem1' and 'problem2'
are used together wit the oligo kernel function (could be extended if you
want to use your own kernel functions).
*/
svm_problem * computeKernelMatrix(const SVMData & problem1, const SVMData & problem2);
/**
@brief This is used for being able to perform predictions with non libsvm standard kernels
*/
void setTrainingSample(svm_problem * training_sample);
/**
@brief This is used for being able to perform predictions with non libsvm standard kernels
*/
void setTrainingSample(SVMData & training_sample);
/**
@brief This function fills probabilities with the probability estimates for the first class.
The libSVM function svm_predict_probability is called to get probability estimates
for the positive class. Since this is only used for binary classification it is sufficient
for every test example to report the probability of the test example belonging to the positive
class. Probability estimates have to be turned on during training (svm.setParameter(PROBABILITY, 1)),
otherwise this method will fill the 'probabilities' vector with -1s.
*/
void getSVCProbabilities(struct svm_problem * problem, std::vector<DoubleReal> & probabilities, std::vector<DoubleReal> & prediction_labels);
/**
@brief Sets weights for the classes in C_SVC (see libsvm documentation for further details)
*/
void setWeights(const std::vector<Int> & weight_labels, const std::vector<DoubleReal> & weights);
private:
/**
@brief find next grid search parameter combination
The current grid cell is given in @p actual_values.
The result is returned in @p actual_values.
*/
bool nextGrid_(const std::vector<DoubleReal> & start_values,
const std::vector<DoubleReal> & step_sizes,
const std::vector<DoubleReal> & end_values,
const bool additive_step_sizes,
std::vector<DoubleReal> & actual_values);
Size getNumberOfEnclosedPoints_(DoubleReal m1, DoubleReal m2, const std::vector<std::pair<DoubleReal, DoubleReal> > & points);
/**
@brief Initializes the svm with standard parameters
*/
void initParameters_();
/**
@brief This function is passed to lib svm for output control
The intention is to discard the output, as we don't need it.
*/
static void printToVoid_(const char * /*s*/);
svm_parameter * param_; // the parameters for the svm
svm_model * model_; // the learnt svm discriminant
DoubleReal sigma_; // for the oligo kernel (amount of positional smearing)
std::vector<DoubleReal> sigmas_; // for the combined oligo kernel (amount of positional smearing)
std::vector<DoubleReal> gauss_table_; // lookup table for fast computation of the oligo kernel
std::vector<std::vector<DoubleReal> > gauss_tables_; // lookup table for fast computation of the combined oligo kernel
Size kernel_type_; // the actual kernel type
Size border_length_; // the actual kernel type
svm_problem * training_set_; // the training set
svm_problem * training_problem_; // the training set
SVMData training_data_; // the training set (different encoding)
};
} // namespace OpenMS
#endif // OPENMS_ANALYSIS_SVM_SVMWRAPPER_H
|