Collects the data related to a single optimization problem. More...

#include <data.h>

Inheritance diagram for dismec::BinaryData:

Public Member Functions
	BinaryData (SparseFeatures x, std::shared_ptr< BinaryLabelVector > y)

long	num_labels () const noexcept override

void	get_labels (label_id_t i, Eigen::Ref< BinaryLabelVector > target) const override

Public Member Functions inherited from dismec::DatasetBase
virtual	~DatasetBase ()=default

	DatasetBase (const DatasetBase &)=default

	DatasetBase (DatasetBase &&)=default

DatasetBase &	operator= (DatasetBase &&)=default

DatasetBase &	operator= (const DatasetBase &)=default

std::shared_ptr< const GenericFeatureMatrix >	get_features () const
	get a shared pointer to the (immutable) feature data More...

std::shared_ptr< GenericFeatureMatrix >	edit_features ()
	get a shared pointer to mutable feature data. Use with care. More...

long	num_features () const noexcept
	Get the total number of features, i.e. the number of columns in the feature matrix. More...

long	num_examples () const noexcept
	Get the total number of instances, i.e. the number of rows in the feature matrix. More...

virtual long	num_positives (label_id_t id) const

virtual long	num_negatives (label_id_t id) const

std::shared_ptr< const BinaryLabelVector >	get_labels (label_id_t id) const

Private Attributes
std::shared_ptr< BinaryLabelVector >	m_Labels

Additional Inherited Members
Protected Member Functions inherited from dismec::DatasetBase
	DatasetBase (SparseFeatures x)

	DatasetBase (DenseFeatures x)

Protected Attributes inherited from dismec::DatasetBase
std::shared_ptr< GenericFeatureMatrix >	m_Features

Detailed Description

Collects the data related to a single optimization problem.

This contains the entire training set as sparse features, so we require that this fit into memory. The lables are strored as a dense vector.

Definition at line 69 of file data.h.

Constructor & Destructor Documentation

◆ BinaryData()

dismec::BinaryData::BinaryData	(	SparseFeatures	x,
		std::shared_ptr< BinaryLabelVector >	y
	)

inline

Definition at line 71 of file data.h.

Member Function Documentation

◆ get_labels()

void BinaryData::get_labels	(	label_id_t	id,
		Eigen::Ref< BinaryLabelVector >	target
	)		const

overridevirtual

Gets the label vector (encoded as dense vector with elements from {-1, 1}) for the id'th class. The weights will be put into the given target buffer. Throws std::out_of_bounds, if id is not in [0, num_labels()).

Implements dismec::DatasetBase.

Definition at line 28 of file data.cpp.

References m_Labels.

◆ num_labels()

long BinaryData::num_labels ( ) const

overridevirtualnoexcept

Gets the total number of different labels in the dataset. TODO call this num_classes instead?

Implements dismec::DatasetBase.

Definition at line 35 of file data.cpp.

Member Data Documentation

◆ m_Labels

std::shared_ptr<BinaryLabelVector> dismec::BinaryData::m_Labels

private

Definition at line 82 of file data.h.

Referenced by get_labels().

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

src/data/data.h
src/data/data.cpp

Public Member Functions

Private Attributes

Additional Inherited Members