mdsEM.h File Reference

Medical Data Segmentation Toolkit (MDSTk)
Copyright (c) 2003-2006 by Michal Spanel
. More...

#include <MDSTk/Base/mdsGlobalLog.h>
#include <MDSTk/Base/mdsHasOptions.h>
#include <MDSTk/Math/mdsRandom.h>
#include <MDSTk/Math/mdsLogNum.h>
#include <MDSTk/Math/mdsMatrix.h>
#include <MDSTk/Math/mdsVector.h>
#include <MDSTk/Math/mdsGaussianFunc.h>
#include <MDSTk/Math/mdsMatrixFunctions.h>
#include <MDSTk/Math/mdsVectorFunctions.h>
#include <vector>
#include <cfloat>
#include "mdsEM.hxx"

Classes
class	mds::math::CMaxLikelihoodByEM< C, N >
	Gaussian Mixture Model (GMM) optimization using maximal likelihood estimation via the Expection-Maximization (EM) algorithm. More...
Namespaces
namespace	mds
	Medical Data Segmentation Toolkit (MDSTk) Copyright (c) 2003-2005 by Michal Spanel
namespace	mds::math
namespace	mds::math::MaxLikelihoodByEM
Enumerations
enum	mds::math::MaxLikelihoodByEM::EOptions { mds::math::MaxLikelihoodByEM::LOGGING_ENABLED = 1 << 0, mds::math::MaxLikelihoodByEM::ADD_RAND_NOISE = 1 << 1, mds::math::MaxLikelihoodByEM::DEFAULT_FLAGS = LOGGING_ENABLED | ADD_RAND_NOISE }
	Flags which may be passed to the EM algorithm. More...
Variables
const tSize	mds::math::MaxLikelihoodByEM::MAX_ITERS = 10
	Maximal allowed number of iterations.
const tSize	mds::math::MaxLikelihoodByEM::MAX_ITERS2 = 10
	Maximal allowed number of internal iterations.
const double	mds::math::MaxLikelihoodByEM::MIN_ADD_CHANGE = 0.001
	Minimal required change of the log-likelihood function which results in a new component addition.
const double	mds::math::MaxLikelihoodByEM::MIN_CHANGE = 1.0e-6
	Minimal required change of the log-likelihood function while iterating.
const double	mds::math::MaxLikelihoodByEM::RAND_NOISE_COEFF = 0.5
	Random noise.
const double	mds::math::MaxLikelihoodByEM::SPLITTING_COEFF = 0.5
	Coefficient used to split a gaussian component.

---

Detailed Description

Medical Data Segmentation Toolkit (MDSTk)
Copyright (c) 2003-2006 by Michal Spanel
.

Authors: Michal Spanel, spanel@fit.vutbr.cz
Date: 2006/08/29

Id:

mdsEM.h 2096 2012-02-16 19:34:46Z spanel

Description:

• Gaussian Mixture Model (GMM) optimization using maximal likelihood estimation via the Expection-Maximization (EM) algorithm.