The RCA Analyzer computes the Relevant Components Analysis of the given DataSet using the specified descriptors. More...

#include <rca.h>

Inheritance diagram for gaia2::RCA:

Public Member Functions
	RCA (const ParameterMap &params)

Transformation	analyze (const DataSet *dataset, const Region &region) const

Public Member Functions inherited from gaia2::Analyzer
	Analyzer (const ParameterMap &params)

virtual Transformation	analyze (const DataSet *dataset) const

void	checkDataSet (const DataSet *dataset) const
	Checks that the given dataset is valid. More...

void	checkMinPoints (const DataSet *dataset, int n) const
	Checks that the given dataset as at least the specified number of points. More...

const Region &	checkFixedLength (const Region &region, const PointLayout &layout) const
	Checks that the given Region only contains fixed-length descriptors and throws an exception if not. More...

Static Public Member Functions
static QList< const PointArray * >	cleanChunklets (QList< PointArray * > &chunklets)

static QList< const PointArray * >	getChunkletsFromMap (const QMap< QString, QString > &filename, const DataSet *inputDataSet)

static QList< const PointArray * >	getChunkletsFromFile (const QString &filename, const DataSet *inputDataSet)

static QList< const PointArray * >	getChunkletsFromLabel (const QString &label, const DataSet *inputDataSet)

Protected Member Functions
Eigen::MatrixXd	computeCovarianceMatrix (const QList< const PointArray * > &chunklets, const Region &region) const
	Computes the within chunklet covariance matrix. More...

Eigen::MatrixXd	whitenMatrix (const Eigen::MatrixXd &cov) const

Protected Attributes
QString	_resultName

int	_targetDimension

QString	_classFile

QString	_className

QMap< QString, QString >	_classMap

Protected Attributes inherited from gaia2::Analyzer
ParameterMap	_params

QStringList	_descriptorNames

QStringList	_exclude

Additional Inherited Members
Public Attributes inherited from gaia2::Analyzer
QString	name
	Name for the algorithm, usually the key that was used to instantiate it from the factory.

QStringList	validParams
	List of valid parameters this analyzer accepts. More...

Detailed Description

The RCA Analyzer computes the Relevant Components Analysis of the given DataSet using the specified descriptors.

Note: the use of classFile, classMap and className is exclusive, ie: you need to specify one and only one of them.

Parameters

descriptorNames	the names of the descriptors to be included in the RCA analysis. Only accepts real descriptors, can use wildcards. (default: "*").
except	the names of the descriptors to be removed from the selected list specified before. Can use wildcards. (default: "").
resultName	the name of the resulting descriptor. (default: some name constructed from 'rca', the number of dimensions, and the original descriptor names)
dimension	the target (final) dimension.
classFile	the name of the file containing the list of classes (chunklets in RCA language).
classMap	the groundtruth map, from point IDs to their class value.
className	the name of the descriptor in the PointLayout that will be used as class label.

Member Function Documentation

Transformation gaia2::RCA::analyze	(	const DataSet *	dataset,
		const Region &	descs
	)		const

virtual

Works exactly like the matlab algo, so it should be ok. However there's something that smells fishy to me so better check it out sometime...

Reimplemented from gaia2::Analyzer.

MatrixXd gaia2::RCA::computeCovarianceMatrix	(	const QList< const PointArray * > &	chunklets,
		const Region &	region
	)		const

protected

Computes the within chunklet covariance matrix.

The implementation follows rigorously the description from the paper, however some doubts subsist (most probably due to my ignorance):

why do we normalize all of them at once, instead of normalizing each one separately?
why do we normalize by n instead of (n-1) (probably not worth much...)

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