Class to extract metrics from either the agent or the environment. More...

#include <mapElitesSelectionMetrics.h>

Inheritance diagram for Selector::MapElites::MapElitesSelectionMetrics:

Public Member Functions
	MapElitesSelectionMetrics (std::vector< std::shared_ptr< const MapElitesDescriptor > > descriptors)
	Default constructor of MapEliteselectionMetrics with a vector of descriptors.

	MapElitesSelectionMetrics (double score, const std::map< std::shared_ptr< const MapElitesDescriptor >, std::vector< double > > &mapDescriptors)
	Constructor of MapEliteselectionMetrics with a vector of scores.

	MapElitesSelectionMetrics (double score)
	Constructor of MapEliteselectionMetrics with a vector of scores.

const std::map< std::shared_ptr< const MapElitesDescriptor >, std::vector< double > > &	getMapDescriptors () const
	Get a const ref to the scorePerClass attribute.

void	initMetrics (const TPG::TPGVertex *agent, const Learn::LearningEnvironment &learningEnvironment) override
	Specialization of the initialisation of the metrics.

void	extractMetricsStep (const TPG::TPGVertex *agent, std::vector< double > actionValues, const Learn::LearningEnvironment &learningEnvironment) override
	Specialization of the extraction of the metrics at the end of an episode.

void	extractMetricsEpisode (const TPG::TPGVertex *agent, size_t nbStepsExecuted, const Learn::LearningEnvironment &learningEnvironment) override
	Specialization of the extraction of the metrics at the end of an episode.

virtual void	weightedSum (std::shared_ptr< SelectionMetrics > other, size_t nbEvaluation, size_t nbEvaluationOther) override
	Specialization of weightedSum method to add the score per class and nbEvalPerClass.

Public Member Functions inherited from Selector::SelectionMetrics
	SelectionMetrics ()=default
	Default constructor.

virtual	~SelectionMetrics ()=default
	Default destructor.

	SelectionMetrics (double score, double utility=0)
	Constructor with score and utility initialization.

virtual double	getScore () const

virtual double	getUtility () const

Protected Attributes
std::map< std::shared_ptr< const MapElitesDescriptor >, std::vector< double > >	mapDescriptors
	Map storing a descriptor and a Vector storing a double value per descriptor (i.e. per Action) of a classification LearningEnvironment.

Protected Attributes inherited from Selector::SelectionMetrics
double	score = 0

double	utility = 0

Additional Inherited Members
Static Public Member Functions inherited from Selector::SelectionMetrics
template<class T >
static T	weightedSum (T value, T valueOther, size_t nbEvaluation, size_t nbEvaluationOther)
	Perform a weighted sum between 2 values.

Detailed Description

Class to extract metrics from either the agent or the environment.

This metrics can be used to specify the selection of the selector. This class does not implement any metrics, it need to be override by the different selection methods.

Constructor & Destructor Documentation

◆ MapElitesSelectionMetrics() [1/3]

Selector::MapElites::MapElitesSelectionMetrics::MapElitesSelectionMetrics ( std::vector< std::shared_ptr< const MapElitesDescriptor > > descriptors )

inline

Default constructor of MapEliteselectionMetrics with a vector of descriptors.

Parameters

[in] descriptors the vector of descriptors to use.

◆ MapElitesSelectionMetrics() [2/3]

Selector::MapElites::MapElitesSelectionMetrics::MapElitesSelectionMetrics	(	double	score,
		const std::map< std::shared_ptr< const MapElitesDescriptor >, std::vector< double > > &	mapDescriptors )

inline

Constructor of MapEliteselectionMetrics with a vector of scores.

Parameters

[in]	score	the score obtained by the agent.
[in]	mapDescriptors	the map of descriptors associated to their descriptor values.

◆ MapElitesSelectionMetrics() [3/3]

Selector::MapElites::MapElitesSelectionMetrics::MapElitesSelectionMetrics ( double score )

inline

Constructor of MapEliteselectionMetrics with a vector of scores.

Parameters

[in] score the score obtained by the agent.

Member Function Documentation

◆ extractMetricsEpisode()

void Selector::MapElites::MapElitesSelectionMetrics::extractMetricsEpisode	(	const TPG::TPGVertex *	agent,
		size_t	nbStepsExecuted,
		const Learn::LearningEnvironment &	learningEnvironment )

overridevirtual

Specialization of the extraction of the metrics at the end of an episode.

Reimplemented from Selector::SelectionMetrics.

◆ extractMetricsStep()

void Selector::MapElites::MapElitesSelectionMetrics::extractMetricsStep	(	const TPG::TPGVertex *	agent,
		std::vector< double >	actionValues,
		const Learn::LearningEnvironment &	learningEnvironment )

overridevirtual

Specialization of the extraction of the metrics at the end of an episode.

Reimplemented from Selector::SelectionMetrics.

◆ initMetrics()

void Selector::MapElites::MapElitesSelectionMetrics::initMetrics	(	const TPG::TPGVertex *	agent,
		const Learn::LearningEnvironment &	learningEnvironment )

overridevirtual

Specialization of the initialisation of the metrics.

Reimplemented from Selector::SelectionMetrics.

◆ weightedSum()

void Selector::MapElites::MapElitesSelectionMetrics::weightedSum	(	std::shared_ptr< SelectionMetrics >	other,
		size_t	nbEvaluation,
		size_t	nbEvaluationOther )

overridevirtual

Specialization of weightedSum method to add the score per class and nbEvalPerClass.

Reimplemented from Selector::SelectionMetrics.

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

D:/a/gegelati/gegelati/gegelatilib/include/selector/mapElites/mapElitesSelectionMetrics.h
D:/a/gegelati/gegelati/gegelatilib/src/selector/mapElites/mapElitesSelectionMetrics.cpp

Public Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ MapElitesSelectionMetrics() [1/3]

◆ MapElitesSelectionMetrics() [2/3]

◆ MapElitesSelectionMetrics() [3/3]

Member Function Documentation

◆ extractMetricsEpisode()

◆ extractMetricsStep()

◆ initMetrics()

◆ weightedSum()