gum::ILearningStrategy Class Reference

<agrum/FMDP/SDyna/ILearningStrategy.h> More...

#include <ILearningStrategy.h>

Inheritance diagram for gum::ILearningStrategy:

Public Member Functions
Constructor & destructor.
virtual	~ILearningStrategy ()
	Destructor (virtual and empty since it's an interface).
Initialization
virtual void	initialize (FMDP< double > *fmdp)=0
	Initializes the learner.
Incremental methods
virtual bool	addObservation (Idx actionId, const Observation *obs)=0
	Gives to the learner a new transition.
virtual void	updateFMDP ()=0
	Starts an update of datastructure in the associated FMDP.
Miscelleanous methods
virtual Size	size ()=0
	learnerSize
virtual const IVisitableGraphLearner *	varLearner (Idx actionId, const DiscreteVariable *var) const =0
	Required for RMax.
virtual double	rMax () const =0
	learnerSize
virtual double	modaMax () const =0
	learnerSize

Detailed Description

<agrum/FMDP/SDyna/ILearningStrategy.h>

Interface for manipulating FMDP learner

Definition at line 74 of file ILearningStrategy.h.

Constructor & Destructor Documentation

~ILearningStrategy()

virtual gum::ILearningStrategy::~ILearningStrategy ( )

inlinevirtual

Destructor (virtual and empty since it's an interface).

Definition at line 84 of file ILearningStrategy.h.

{}

Member Function Documentation

addObservation()

virtual bool gum::ILearningStrategy::addObservation ( Idx actionId, const Observation * obs )

pure virtual

Gives to the learner a new transition.

Parameters

actionId	: the action on which the transition was made
obs	: the observed transition

Returns

true if learning this transition implies structural changes (can trigger a new planning)

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

initialize()

virtual void gum::ILearningStrategy::initialize ( FMDP< double > * fmdp )

pure virtual

Initializes the learner.

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

modaMax()

virtual double gum::ILearningStrategy::modaMax ( ) const

pure virtual

learnerSize

Returns

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

rMax()

virtual double gum::ILearningStrategy::rMax ( ) const

pure virtual

learnerSize

Returns

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

size()

virtual Size gum::ILearningStrategy::size ( )

pure virtual

learnerSize

Returns

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

updateFMDP()

virtual void gum::ILearningStrategy::updateFMDP ( )

pure virtual

Starts an update of datastructure in the associated FMDP.

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

varLearner()

virtual const IVisitableGraphLearner * gum::ILearningStrategy::varLearner ( Idx actionId, const DiscreteVariable * var ) const

pure virtual

Required for RMax.

Returns

Implemented in gum::FMDPLearner< VariableAttributeSelection, RewardAttributeSelection, LearnerSelection >.

---

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

• agrum/FMDP/SDyna/Strategies/ILearningStrategy.h