trainer.h File Reference

#include <vector>
#include <dlib/matrix.h>
#include <dlib/svm.h>

Include dependency graph for trainer.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
class	LDATrainer

Functions
static void	GramSchmidt (dlib::matrix< LDATrainer::sample_type, 0, 1 > &basis, LDATrainer::sample_type &newX)
	Gram-Schmidt process to re-orthonormalise the basis. More...
static void	ComputeReferencePoints (Classifier::Point2D &refpt_pos, Classifier::Point2D &refpt_neg, const std::vector< float > &proj1, const std::vector< float > &proj2, const std::vector< float > &labels, unsigned *_npos=0, unsigned *_nneg=0)
	Compute pos. and neg. reference points. More...
static bool	DilateClassifier (Classifier &classifier, std::vector< float > &proj1, std::vector< float > &proj2, const std::vector< float > &labels, const std::vector< LDATrainer::sample_type > &samples, LDATrainer &trainer, LDATrainer &orthoTrainer)

Function Documentation

ComputeReferencePoints()

static void ComputeReferencePoints ( Classifier::Point2D &  refpt_pos, Classifier::Point2D &  refpt_neg, const std::vector< float > &  proj1, const std::vector< float > &  proj2, const std::vector< float > &  labels, unsigned *  _npos = 0, unsigned *  _nneg = 0  )

static

Compute pos. and neg. reference points.

Definition at line 248 of file trainer.h.

Referenced by DilateClassifier(), and qCanupoTools::TrainClassifier().

DilateClassifier()

static bool DilateClassifier ( Classifier &  classifier, std::vector< float > &  proj1, std::vector< float > &  proj2, const std::vector< float > &  labels, const std::vector< LDATrainer::sample_type > &  samples, LDATrainer &  trainer, LDATrainer &  orthoTrainer  )

static

Experimental (same as Brodu's code): dilatation to highlight the internal data structure

Definition at line 279 of file trainer.h.

References Classifier::axisScaleRatio, ComputeReferencePoints(), Vector2Tpl< Type >::dot(), LDATrainer::m_weights, Vector2Tpl< Type >::normalize(), LDATrainer::predict(), Classifier::refPointNeg, Classifier::refPointPos, Classifier::weightsAxis1, Classifier::weightsAxis2, Vector2Tpl< Type >::x, and Vector2Tpl< Type >::y.

Referenced by qCanupoTools::TrainClassifier().

GramSchmidt()

static void GramSchmidt ( dlib::matrix< LDATrainer::sample_type, 0, 1 > &  basis, LDATrainer::sample_type &  newX  )

static

Gram-Schmidt process to re-orthonormalise the basis.

Definition at line 215 of file trainer.h.

References dot(), and fabs().

Referenced by qCanupoTools::TrainClassifier().