Fitter object that encapsulates density based classification using instances of ModelFittingDensityEstimation for each class. More...

#include <ModelFittingClassification.hpp>

Inheritance diagram for sgpp::datadriven::ModelFittingClassification:

Public Member Functions
bool	adapt () override
	Improve the accuracy of the classification by refining or coarsening the grids of each class.

double	computeResidual (DataMatrix &validationData) const override
	Should compute some kind of Residual to evaluate the fit of the model.

double	evaluate (const DataVector &sample) override
	Predict the class of a data sample based on the density of the sample for each model.

void	evaluate (DataMatrix &samples, DataVector &results) override
	Predicts the class for a set of data points based on the learned densities for each class.

void	fit (Dataset &dataset) override
	Fits the models for all classes based on the data given in the dataset parameter.

void	fit (Dataset &datasetP, Dataset &datasetQ) override

std::map< double, size_t >	getClassIdx ()
	obtain the index mapping for each label class.

std::vector< std::unique_ptr< ModelFittingDensityEstimation > > *	getModels ()
	obtain the density estimation models per each class.

	ModelFittingClassification (const FitterConfigurationClassification &config)
	Constructor.

	ModelFittingClassification (const FitterConfigurationClassification &config, std::shared_ptr< DBMatObjectStore > objectStore)
	Constructor with specified object store.

void	reset () override
	Resets the state of the entire model.

void	resetTraining () override
	Resets any trained representations of the model, but does not reset the entire state.

void	storeClassificator ()
	store Fitter into text file in folder /datadriven/classificator/

void	update (Dataset &dataset) override
	Updates the models for each class based on new data (streaming or batch learning)

void	update (Dataset &datasetP, Dataset &datasetQ) override

void	updateRegularization (double lambda) override
	Updates the regularization parameter lambda of the underlying model.

Public Member Functions inherited from sgpp::datadriven::ModelFittingBase
Dataset *	getDataset ()

FitterConfiguration &	getFitterConfiguration ()
	Get or set the configuration of the fitter object.

const FitterConfiguration &	getFitterConfiguration () const
	Get the configuration of the fitter object.

virtual Grid &	getGrid ()
	Return the learned grid.

virtual std::shared_ptr< BlacsProcessGrid >	getProcessGrid () const

virtual DataVector &	getSurpluses ()
	Return the learned hierarchical surpluses.

	ModelFittingBase ()
	Default constructor.

	ModelFittingBase (const ModelFittingBase &rhs)=delete
	Copy constructor - we cannot deep copy all member variables yet.

	ModelFittingBase (ModelFittingBase &&rhs)=default
	Move constructor.

ModelFittingBase &	operator= (const ModelFittingBase &rhs)=delete
	Copy assign operator - we cannot deep copy all member variables yet.

ModelFittingBase &	operator= (ModelFittingBase &&rhs)=default
	Move assign operator.

virtual	~ModelFittingBase ()=default
	virtual destructor.

Additional Inherited Members
Public Attributes inherited from sgpp::datadriven::ModelFittingBase
bool	verboseSolver
	Whether the Solver produces output or not.

Protected Member Functions inherited from sgpp::datadriven::ModelFittingBase
Grid *	buildGrid (const sgpp::base::GeneralGridConfiguration &gridConfig) const
	Factory member function that generates a grid from configuration.

Grid *	buildGrid (const sgpp::base::GeneralGridConfiguration &gridConfig, const GeometryConfiguration &geometryConfig) const
	Factory member function that generates a grid from configuration.

SLESolver *	buildSolver (const SLESolverConfiguration &config) const
	Factory member function to build the solver for the least squares regression problem according to the config.

std::set< std::set< size_t > >	getInteractions (const GeometryConfiguration &geometryConfig)

void	reconfigureSolver (SLESolver &solver, const SLESolverConfiguration &config) const
	Configure solver based on the desired configuration.

Protected Attributes inherited from sgpp::datadriven::ModelFittingBase
std::unique_ptr< FitterConfiguration >	config
	Configuration object for the fitter.

Dataset *	dataset
	Pointer to sgpp::datadriven::Dataset.

Dataset *	extraDataset

std::unique_ptr< std::set< std::set< size_t > > >	interactions

std::unique_ptr< SLESolver >	solver
	Solver for the learning problem.

Detailed Description

Fitter object that encapsulates density based classification using instances of ModelFittingDensityEstimation for each class.

Constructor & Destructor Documentation

◆ ModelFittingClassification() [1/2]

sgpp::datadriven::ModelFittingClassification::ModelFittingClassification ( const FitterConfigurationClassification & config )

explicit

Constructor.

Parameters

config configuration object that specifies grid, refinement, and regularization

References sgpp::datadriven::ModelFittingBase::config, and sgpp::datadriven::FitterConfiguration::getParallelConfig().

◆ ModelFittingClassification() [2/2]

sgpp::datadriven::ModelFittingClassification::ModelFittingClassification	(	const FitterConfigurationClassification &	config,
		std::shared_ptr< DBMatObjectStore >	objectStore
	)

explicit

Constructor with specified object store.

Parameters

config	Configuration object that specifies grid, refinement, and regularization
objectStore	Offline object store for already decomposed offline objects.

Member Function Documentation

◆ adapt()

bool sgpp::datadriven::ModelFittingClassification::adapt ( )

overridevirtual

Improve the accuracy of the classification by refining or coarsening the grids of each class.

Coarsening is currently only implemented for RefinementFunctorType::Classification

Returns: true if refinement could be performed for any grid based on the refinement configuration, else false.

Implements sgpp::datadriven::ModelFittingBase.

References sgpp::datadriven::ClassificationRefinementFunctor::adaptAllGrids(), sgpp::base::ComponentGrid, sgpp::datadriven::ModelFittingBase::config, sgpp::datadriven::ModelFittingBase::getGrid(), sgpp::datadriven::GridFactory::getInteractions(), sgpp::datadriven::ModelFittingBase::getSurpluses(), sgpp::base::AdaptivityConfiguration::numRefinements_, sgpp::datadriven::MultiGridRefinementFunctor::preComputeEvaluations(), sgpp::base::AdaptivityConfiguration::precomputeEvaluations_, sgpp::datadriven::MultipleClassRefinementFunctor::refine(), sgpp::base::AdaptivityConfiguration::refinementFunctorType_, sgpp::datadriven::MultiGridRefinementFunctor::setGridIndex(), and sgpp::datadriven::GeometryConfiguration::stencils_.

◆ computeResidual()

double sgpp::datadriven::ModelFittingClassification::computeResidual ( DataMatrix & validationData ) const

inlineoverridevirtual

Should compute some kind of Residual to evaluate the fit of the model.

In the case of density estimation, this is || R * alpha_lambda - b_val ||_2

This is useful for unsupervised learning models, where normal evaluation cannot be used as there are no targets.

For classification, this is not implemented, as accuracy should be used in this case.

Parameters

validationData Matrix for validation data

Returns: the residual score

Implements sgpp::datadriven::ModelFittingBase.

◆ evaluate() [1/2]

double sgpp::datadriven::ModelFittingClassification::evaluate ( const DataVector & sample )

overridevirtual

Predict the class of a data sample based on the density of the sample for each model.

Parameters

sample the sample point to classify

Returns: the predicted class label

Implements sgpp::datadriven::ModelFittingBase.

Referenced by evaluate(), sgpp::datadriven::VisualizerClassification::storeHeatmapJsonClassification(), and sgpp::datadriven::VisualizerClassification::storeHeatmapJsonClassification().

◆ evaluate() [2/2]

void sgpp::datadriven::ModelFittingClassification::evaluate	(	DataMatrix &	samples,
		DataVector &	results
	)

overridevirtual

Predicts the class for a set of data points based on the learned densities for each class.

Parameters

samples	matrix where each row represents a data sample
results	vector to output the predicted classes

Implements sgpp::datadriven::ModelFittingBase.

References sgpp::datadriven::ModelFittingBase::config, evaluate(), sgpp::datadriven::DataVectorDistributed::getLocalPointer(), sgpp::datadriven::DataVectorDistributed::getLocalRows(), sgpp::datadriven::DataVectorDistributed::localToGlobalRowIndex(), sgpp::base::DataVector::mult(), sgpp::base::DataVector::set(), and sgpp::datadriven::DataVectorDistributed::toLocalDataVector().

◆ fit() [1/2]

void sgpp::datadriven::ModelFittingClassification::fit ( Dataset & dataset )

overridevirtual

Fits the models for all classes based on the data given in the dataset parameter.

Parameters

dataset the training dataset that is used to fit the models

Implements sgpp::datadriven::ModelFittingBase.

References reset(), and update().

◆ fit() [2/2]

void sgpp::datadriven::ModelFittingClassification::fit	(	Dataset &	datasetP,
		Dataset &	datasetQ
	)

inlineoverridevirtual

Implements sgpp::datadriven::ModelFittingBase.

◆ getClassIdx()

std::map< double, size_t > sgpp::datadriven::ModelFittingClassification::getClassIdx ( )

obtain the index mapping for each label class.

To be used in VisualizerClassification

Referenced by sgpp::datadriven::VisualizerClassification::runVisualization().

◆ getModels()

std::vector< std::unique_ptr< ModelFittingDensityEstimation > > * sgpp::datadriven::ModelFittingClassification::getModels ( )

obtain the density estimation models per each class.

To be used in VisualizerClassification

Referenced by sgpp::datadriven::VisualizerClassification::runVisualization(), sgpp::datadriven::VisualizerClassification::storeHeatmapJsonClassification(), and sgpp::datadriven::VisualizerClassification::storeHeatmapJsonClassification().

◆ reset()

void sgpp::datadriven::ModelFittingClassification::reset ( )

overridevirtual

Resets the state of the entire model.

Implements sgpp::datadriven::ModelFittingBase.

Referenced by fit().

◆ resetTraining()

void sgpp::datadriven::ModelFittingClassification::resetTraining ( )

overridevirtual

Resets any trained representations of the model, but does not reset the entire state.

Decompositions are not discarded, but can be reused.

Implements sgpp::datadriven::ModelFittingBase.

◆ storeClassificator()

void sgpp::datadriven::ModelFittingClassification::storeClassificator ( )

store Fitter into text file in folder /datadriven/classificator/

◆ update() [1/2]

void sgpp::datadriven::ModelFittingClassification::update ( Dataset & dataset )

overridevirtual

Updates the models for each class based on new data (streaming or batch learning)

Parameters

dataset the new data

Implements sgpp::datadriven::ModelFittingBase.

References sgpp::datadriven::ModelFittingBase::dataset, sgpp::base::DataVector::get(), sgpp::datadriven::Dataset::getData(), sgpp::datadriven::Dataset::getDimension(), sgpp::datadriven::Dataset::getNumberInstances(), sgpp::base::DataMatrix::getRow(), and sgpp::datadriven::Dataset::getTargets().

Referenced by fit(), python.uq.refinement.RefinementStrategy.Ranking::rank(), and python.learner.LearnedKnowledge.LearnedKnowledge::setMemento().

◆ update() [2/2]

void sgpp::datadriven::ModelFittingClassification::update	(	Dataset &	datasetP,
		Dataset &	datasetQ
	)

inlineoverridevirtual

Implements sgpp::datadriven::ModelFittingBase.

Referenced by python.uq.refinement.RefinementStrategy.Ranking::rank(), and python.learner.LearnedKnowledge.LearnedKnowledge::setMemento().

◆ updateRegularization()

void sgpp::datadriven::ModelFittingClassification::updateRegularization ( double lambda )

overridevirtual

Updates the regularization parameter lambda of the underlying model.

Parameters

lambda the new lambda parameter

Implements sgpp::datadriven::ModelFittingBase.

References lambda.

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

datadriven/src/sgpp/datadriven/datamining/modules/fitting/ModelFittingClassification.hpp
datadriven/src/sgpp/datadriven/datamining/modules/fitting/ModelFittingClassification.cpp

Public Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ ModelFittingClassification() [1/2]

◆ ModelFittingClassification() [2/2]

Member Function Documentation

◆ adapt()

◆ computeResidual()

◆ evaluate() [1/2]

◆ evaluate() [2/2]

◆ fit() [1/2]

◆ fit() [2/2]

◆ getClassIdx()

◆ getModels()

◆ reset()

◆ resetTraining()

◆ storeClassificator()

◆ update() [1/2]

◆ update() [2/2]

◆ updateRegularization()