Leave-one-out error objective function. More...

#include <shark/ObjectiveFunctions/LooError.h>

Inheritance diagram for shark::LooError< ModelTypeT, LabelType >:

Public Types
typedef ModelTypeT	ModelType

typedef ModelType::InputType	InputType

typedef ModelType::OutputType	OutputType

typedef LabeledData< InputType, LabelType >	DatasetType

typedef AbstractTrainer< ModelType, LabelType >	TrainerType

typedef AbstractLoss< LabelType, typename ModelType::OutputType >	LossType

Public Types inherited from shark::AbstractObjectiveFunction< RealVector, double >
enum	Feature
	List of features that are supported by an implementation. More...

typedef RealVector	SearchPointType

typedef double	ResultType

typedef boost::mpl::if_< std::is_arithmetic< double >, SearchPointType, RealMatrix >::type	FirstOrderDerivative

typedef TypedFlags< Feature >	Features
	This statement declares the member m_features. See Core/Flags.h for details.

typedef TypedFeatureNotAvailableException< Feature >	FeatureNotAvailableException

Public Member Functions
	LooError (DatasetType const &dataset, ModelType model, TrainerType trainer, LossType loss, IParameterizable<> meta=NULL)
	Constructor.

std::string	name () const
	From INameable: return the class name.

std::size_t	numberOfVariables () const
	Accesses the number of variables.

double	eval () const

double	eval (const RealVector &parameters) const

Public Member Functions inherited from shark::AbstractObjectiveFunction< RealVector, double >
const Features &	features () const

virtual void	updateFeatures ()

bool	hasValue () const
	returns whether this function can calculate it's function value

bool	hasFirstDerivative () const
	returns whether this function can calculate the first derivative

bool	hasSecondDerivative () const
	returns whether this function can calculate the second derivative

bool	canProposeStartingPoint () const
	returns whether this function can propose a starting point.

bool	isConstrained () const
	returns whether this function can return

bool	hasConstraintHandler () const
	returns whether this function can return

bool	canProvideClosestFeasible () const
	Returns whether this function can calculate thee closest feasible to an infeasible point.

bool	isThreadSafe () const
	Returns true, when the function can be usd in parallel threads.

bool	isNoisy () const
	Returns true, when the function can be usd in parallel threads.

	AbstractObjectiveFunction ()
	Default ctor.

virtual	~AbstractObjectiveFunction ()
	Virtual destructor.

virtual void	init ()

void	setRng (random::rng_type *rng)
	Sets the Rng used by the objective function.

virtual bool	hasScalableDimensionality () const

virtual void	setNumberOfVariables (std::size_t numberOfVariables)
	Adjusts the number of variables if the function is scalable.

virtual std::size_t	numberOfObjectives () const

virtual bool	hasScalableObjectives () const

virtual void	setNumberOfObjectives (std::size_t numberOfObjectives)
	Adjusts the number of objectives if the function is scalable.

std::size_t	evaluationCounter () const
	Accesses the evaluation counter of the function.

AbstractConstraintHandler< SearchPointType > const &	getConstraintHandler () const
	Returns the constraint handler of the function if it has one.

virtual bool	isFeasible (const SearchPointType &input) const
	Tests whether a point in SearchSpace is feasible, e.g., whether the constraints are fulfilled.

virtual void	closestFeasible (SearchPointType &input) const
	If supported, the supplied point is repaired such that it satisfies all of the function's constraints.

virtual SearchPointType	proposeStartingPoint () const
	Proposes a starting point in the feasible search space of the function.

ResultType	operator() (SearchPointType const &input) const
	Evaluates the function. Useful together with STL-Algorithms like std::transform.

virtual ResultType	evalDerivative (SearchPointType const &input, FirstOrderDerivative &derivative) const
	Evaluates the objective function and calculates its gradient.

virtual ResultType	evalDerivative (SearchPointType const &input, SecondOrderDerivative &derivative) const
	Evaluates the objective function and calculates its gradient.

Public Member Functions inherited from shark::INameable
virtual	~INameable ()

Protected Attributes
DataView< DatasetType const >	m_dataset

IParameterizable *	mep_meta

ModelType *	mep_model

TrainerType *	mep_trainer

LossType *	mep_loss

Protected Attributes inherited from shark::AbstractObjectiveFunction< RealVector, double >
Features	m_features

std::size_t	m_evaluationCounter
	Evaluation counter, default value: 0.

AbstractConstraintHandler< SearchPointType > const *	m_constraintHandler

random::rng_type *	mep_rng

Additional Inherited Members
Protected Member Functions inherited from shark::AbstractObjectiveFunction< RealVector, double >
void	announceConstraintHandler (AbstractConstraintHandler< SearchPointType > const *handler)
	helper function which is called to announce the presence of an constraint handler.

Detailed Description

template<class ModelTypeT, class LabelType = typename ModelTypeT::OutputType>
class shark::LooError< ModelTypeT, LabelType >

Leave-one-out error objective function.

: The leave-one-out measure is the average prediction performance of a learning machine on a dataset, where each sample is predicted by a machine trained on all but the sample to be predicted. This is an extreme form of cross-validation, with a fold size of one.

: In general the leave-one-out error is costly to compute, since it requires training of a large number of learning machines. However, certain machines allow for a more efficient implementation. Refer to LooErrorCSvm for an example.

Definition at line 63 of file LooError.h.

Member Typedef Documentation

◆ DatasetType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef LabeledData<InputType, LabelType> shark::LooError< ModelTypeT, LabelType >::DatasetType

Definition at line 69 of file LooError.h.

◆ InputType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef ModelType::InputType shark::LooError< ModelTypeT, LabelType >::InputType

Definition at line 67 of file LooError.h.

◆ LossType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef AbstractLoss<LabelType, typename ModelType::OutputType> shark::LooError< ModelTypeT, LabelType >::LossType

Definition at line 71 of file LooError.h.

◆ ModelType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef ModelTypeT shark::LooError< ModelTypeT, LabelType >::ModelType

Definition at line 66 of file LooError.h.

◆ OutputType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef ModelType::OutputType shark::LooError< ModelTypeT, LabelType >::OutputType

Definition at line 68 of file LooError.h.

◆ TrainerType

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

typedef AbstractTrainer<ModelType, LabelType> shark::LooError< ModelTypeT, LabelType >::TrainerType

Definition at line 70 of file LooError.h.

Constructor & Destructor Documentation

◆ LooError()

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

shark::LooError< ModelTypeT, LabelType >::LooError	(	DatasetType const &	dataset,
		ModelType *	model,
		TrainerType *	trainer,
		LossType *	loss,
		IParameterizable<> *	meta = `NULL`
	)

inline

Constructor.

Parameters

dataset	Full data set for leave-one-out.
model	Model built on subsets of the data.
trainer	Trainer for learning on each subset.
loss	Loss function for judging the validation output.
meta	Meta object with parameters that influences the process, typically a trainer.

Definition at line 82 of file LooError.h.

References shark::AbstractObjectiveFunction< RealVector, double >::HAS_VALUE, and shark::AbstractObjectiveFunction< RealVector, double >::m_features.

Member Function Documentation

◆ eval() [1/2]

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

double shark::LooError< ModelTypeT, LabelType >::eval ( ) const

inline

Evaluate the leave-one-out error: train sub-models, evaluate objective, return the average.

Definition at line 114 of file LooError.h.

References shark::AbstractLoss< LabelT, OutputT >::eval(), shark::LooError< ModelTypeT, LabelType >::m_dataset, shark::AbstractObjectiveFunction< RealVector, double >::m_evaluationCounter, shark::LooError< ModelTypeT, LabelType >::mep_loss, shark::LooError< ModelTypeT, LabelType >::mep_model, shark::LooError< ModelTypeT, LabelType >::mep_trainer, shark::subset(), shark::toDataset(), and shark::AbstractTrainer< Model, LabelTypeT >::train().

Referenced by shark::LooError< ModelTypeT, LabelType >::eval().

◆ eval() [2/2]

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

double shark::LooError< ModelTypeT, LabelType >::eval ( const RealVector & parameters ) const

inlinevirtual

Evaluate the leave-one-out error for the given parameters passed to the meta object (typically these parameters need to be optimized in a model selection procedure).

Reimplemented from shark::AbstractObjectiveFunction< RealVector, double >.

Definition at line 138 of file LooError.h.

References shark::LooError< ModelTypeT, LabelType >::eval(), shark::LooError< ModelTypeT, LabelType >::mep_meta, shark::IParameterizable< VectorType >::setParameterVector(), and SHARK_ASSERT.

◆ name()

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

std::string shark::LooError< ModelTypeT, LabelType >::name ( ) const

inlinevirtual

From INameable: return the class name.

Reimplemented from shark::INameable.

Definition at line 99 of file LooError.h.

References shark::LooError< ModelTypeT, LabelType >::mep_loss, shark::LooError< ModelTypeT, LabelType >::mep_model, shark::LooError< ModelTypeT, LabelType >::mep_trainer, and shark::INameable::name().

◆ numberOfVariables()

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

std::size_t shark::LooError< ModelTypeT, LabelType >::numberOfVariables ( ) const

inlinevirtual

Accesses the number of variables.

Implements shark::AbstractObjectiveFunction< RealVector, double >.

Definition at line 107 of file LooError.h.

References shark::LooError< ModelTypeT, LabelType >::mep_meta, and shark::IParameterizable< VectorType >::numberOfParameters().

Member Data Documentation

◆ m_dataset

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

DataView<DatasetType const> shark::LooError< ModelTypeT, LabelType >::m_dataset

protected

Definition at line 144 of file LooError.h.

Referenced by shark::LooError< ModelTypeT, LabelType >::eval().

◆ mep_loss

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

LossType* shark::LooError< ModelTypeT, LabelType >::mep_loss

protected

Definition at line 148 of file LooError.h.

Referenced by shark::LooError< ModelTypeT, LabelType >::eval(), and shark::LooError< ModelTypeT, LabelType >::name().

◆ mep_meta

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

IParameterizable* shark::LooError< ModelTypeT, LabelType >::mep_meta

protected

Definition at line 145 of file LooError.h.

Referenced by shark::LooError< ModelTypeT, LabelType >::eval(), and shark::LooError< ModelTypeT, LabelType >::numberOfVariables().

◆ mep_model

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

ModelType* shark::LooError< ModelTypeT, LabelType >::mep_model

protected

Definition at line 146 of file LooError.h.

Referenced by shark::LooError< ModelTypeT, LabelType >::eval(), and shark::LooError< ModelTypeT, LabelType >::name().

◆ mep_trainer

template<class ModelTypeT , class LabelType = typename ModelTypeT::OutputType>

TrainerType* shark::LooError< ModelTypeT, LabelType >::mep_trainer

protected

Definition at line 147 of file LooError.h.

Referenced by shark::LooError< ModelTypeT, LabelType >::eval(), and shark::LooError< ModelTypeT, LabelType >::name().

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

include/shark/ObjectiveFunctions/LooError.h

Public Types

Public Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ DatasetType

◆ InputType

◆ LossType

◆ ModelType

◆ OutputType

◆ TrainerType

Constructor & Destructor Documentation

◆ LooError()

Member Function Documentation

◆ eval() [1/2]

◆ eval() [2/2]

◆ name()

◆ numberOfVariables()

Member Data Documentation

◆ m_dataset

◆ mep_loss

◆ mep_meta

◆ mep_model

◆ mep_trainer