lti::rbf::initFunctor1 Class Reference

initialization functor for rbf More...

#include <ltiRbf.h>

Inheritance diagram for lti::rbf::initFunctor1:

[legend]

Collaboration diagram for lti::rbf::initFunctor1:

[legend]

List of all members.

Public Member Functions
	initFunctor1 (const dmatrix &theFeatures, const ivector &theTrainIDs, const ivector &theNbViewsObj, const int &theNbObj, const int &theNbNeurObj, const parameters::eLvqInit &theLvqInit, const eNormType &norm)
	~initFunctor1 ()
bool	operator() (dmatrix &weights, ivector &outID)
void	setTrainParams (const parameters::eLvqTrainType &theTrainType, const double &theLearnRate, const double &theLrFac=0., const double &theWinSize=0.)
Protected Attributes
const dmatrix &	features
const ivector &	trainIDs
const ivector &	nbViewsObj
int	nbObj
int	nbNeurObj
double	learnRate
double	lrFac
double	winSize
bool	modified

---

Detailed Description

initialization functor for rbf

---

Constructor & Destructor Documentation

lti::rbf::initFunctor1::initFunctor1	(	const dmatrix &	theFeatures,
		const ivector &	theTrainIDs,
		const ivector &	theNbViewsObj,
		const int &	theNbObj,
		const int &	theNbNeurObj,
		const parameters::eLvqInit &	theLvqInit,
		const eNormType &	norm
	)

default constructor

Parameters:

	theFeatures	the matrix with the training patterns (each row a pattern)
	theTrainIDs	the numerical ids for each pattern in theFeatures
	theNbViewsObj	the number of patterns per class
	theNbObj	the number of classes (or objects)
	theNbNeurObj	the number of neurons per class
	theLvqInit	initialization method vor the LVQ training
	norm	the used distance (L1distance or L2distance)

lti::rbf::initFunctor1::~initFunctor1

(

)

destructor

---

Member Function Documentation

bool lti::rbf::initFunctor1::operator()	(	dmatrix &	weights,
		ivector &	outID
	)			[virtual]

apply operator

Implements lti::stdLayer::initFunctor.

void lti::rbf::initFunctor1::setTrainParams	(	const parameters::eLvqTrainType &	theTrainType,
		const double &	theLearnRate,
		const double &	theLrFac = 0.,
		const double &	theWinSize = 0.
	)

set the train parameters

Parameters:

	theTrainType	specify training for the LVQ initialization
	theLearnRate
	theLrFac	the learn factor
	theWinSize	the window size

---

Member Data Documentation

const dmatrix& lti::rbf::initFunctor1::features [protected]

a reference to the training patterns

double lti::rbf::initFunctor1::learnRate [protected]

learn rate

double lti::rbf::initFunctor1::lrFac [protected]

learn factor

bool lti::rbf::initFunctor1::modified [protected]

flag that indicates whether the weights were changed at all during the last epoche.

If not no further 'training' is performed.

int lti::rbf::initFunctor1::nbNeurObj [protected]

number of neurons per object

int lti::rbf::initFunctor1::nbObj [protected]

number of objects

const ivector& lti::rbf::initFunctor1::nbViewsObj [protected]

a reference to the number of views per object

const ivector& lti::rbf::initFunctor1::trainIDs [protected]

a reference to the numerical ids of the training patterns

double lti::rbf::initFunctor1::winSize [protected]

window size

---

The documentation for this class was generated from the following file:

• ltiRbf.h