gum::InfluenceDiagramInference< GUM_SCALAR > Class Template Reference

<agrum/ID/inference/influenceDiagramInference.h> More...

#include <influenceDiagramInference.h>

Inheritance diagram for gum::InfluenceDiagramInference< GUM_SCALAR >:

Collaboration diagram for gum::InfluenceDiagramInference< GUM_SCALAR >:

Public Member Functions
const InfluenceDiagram< GUM_SCALAR > &	influenceDiagram () const
	Returns a constant reference over the InfluenceDiagram on which this class work. More...
Constructor & destructor
	InfluenceDiagramInference (const InfluenceDiagram< GUM_SCALAR > &infDiag)
	Default constructor. More...
virtual	~InfluenceDiagramInference ()
	Destructor. More...
Inference
virtual void	makeInference ()
GUM_SCALAR	getMEU ()
Idx	getBestDecisionChoice (NodeId decisionId)
std::string	displayResult ()
	displays the result of an inference More...
Evidence management
virtual void	insertEvidence (const List< const Potential< GUM_SCALAR > * > &evidenceList)
virtual void	eraseEvidence (const Potential< GUM_SCALAR > *evidence)
virtual void	eraseAllEvidence ()
Getters & setters
Triangulation &	getTriangulation ()
	Returns the Triangulation used by this class. More...
void	displayStrongJunctionTree (std::ostream &stream=std::cout)
	Displays on terminal the result of strong junction tree computation for test purpose only. More...

Protected Attributes
const InfluenceDiagram< GUM_SCALAR > &	__infDiag
	The Bayes net we wish to perform inference on. More...

Detailed Description

template<typename GUM_SCALAR>
class gum::InfluenceDiagramInference< GUM_SCALAR >

<agrum/ID/inference/influenceDiagramInference.h>

This class implements an algorithm for inference in influence diagrams based upon Shaffer-Shenoy's one for bayes net inferences

The class used for the triangulation is partialOrderedTriangulation.

Definition at line 64 of file influenceDiagramInference.h.

Constructor & Destructor Documentation

InfluenceDiagramInference()

template<typename GUM_SCALAR >

gum::InfluenceDiagramInference< GUM_SCALAR >::InfluenceDiagramInference ( const InfluenceDiagram< GUM_SCALAR > &  infDiag )

explicit

Default constructor.

Parameters

infDiag

the influence diagram we want to perform inference upon

~InfluenceDiagramInference()

template<typename GUM_SCALAR >

virtual gum::InfluenceDiagramInference< GUM_SCALAR >::~InfluenceDiagramInference ( )

virtual

Destructor.

Member Function Documentation

__absorbClique()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__absorbClique ( NodeId  absorbedCliqueId, NodeId  absorbingCliqueId  )

private

Performs the operation of absorption of a clique by another.

__cleanUp()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__cleanUp ( )

private

Cleans Up remaining stuff due to inference.

__collectChild()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__collectChild ( NodeId  parent, NodeId  child  )

private

collect child clique for inferences

__getClique()

template<typename GUM_SCALAR >

NodeId gum::InfluenceDiagramInference< GUM_SCALAR >::__getClique ( const std::vector< NodeId > &  eliminationOrder, NodeId  id  )

private

Returns

Returns the clique in which the node's potentials must be stored

__getSeparator()

template<typename GUM_SCALAR >

const NodeSet& gum::InfluenceDiagramInference< GUM_SCALAR >::__getSeparator ( NodeId  clique_1, NodeId  clique_2  )

private

Returns

Returns a separator given two adjacent cliques

__IsEliminatedAfter()

template<typename GUM_SCALAR >

bool gum::InfluenceDiagramInference< GUM_SCALAR >::__IsEliminatedAfter ( NodeId  observedNode, NodeId  currentNode  )

private

Returns true if observed node is eliminated after current node.

__makeCliquePropertiesMap()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__makeCliquePropertiesMap ( )

private

Builds the cliques tables Uses __getCliquesTable to initialize the cliques table, and multiply the tables with the adequate CPT.

__makeDummyPotential()

template<typename GUM_SCALAR >

Potential< GUM_SCALAR >* gum::InfluenceDiagramInference< GUM_SCALAR >::__makeDummyPotential ( NodeId  cliqueId )

private

Returns a pointer over a "dummy" potential, which is a CPT filled with one MultiDimSparse filled with 1. This is used by empty cliques.

Parameters

cliqueId

The NodeId of the cliqueId for which we build a dummy potential.

Returns

A pointer over the dummy bucket.

__makeDummyUtility()

template<typename GUM_SCALAR >

Potential< GUM_SCALAR >* gum::InfluenceDiagramInference< GUM_SCALAR >::__makeDummyUtility ( NodeId  cliqueId )

private

Returns a pointer over a "dummy" utility, which is a utility table filled with one MultiDimSparse filled with 0. This is used by empty cliques.

Parameters

cliqueId

The NodeId of the cliqueId for which we build a dummy utility.

Returns

A pointer over the dummy bucket.

__makeStrongJunctionTree()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__makeStrongJunctionTree ( )

private

Makes a strong junction tree that make easier elimination.

__reduceClique()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::__reduceClique ( CliqueProperties< GUM_SCALAR > *  absorbedClique, NodeSet &  separator, Potential< GUM_SCALAR > *&  potentialMarginal, Potential< GUM_SCALAR > *&  utilityMarginal  )

private

Reduces a clique down to her separator from another clique elements.

displayResult()

template<typename GUM_SCALAR >

std::string gum::InfluenceDiagramInference< GUM_SCALAR >::displayResult

(

)

displays the result of an inference

displayStrongJunctionTree()

template<typename GUM_SCALAR >

void gum::InfluenceDiagramInference< GUM_SCALAR >::displayStrongJunctionTree

(

std::ostream &

stream = std::cout

)

Displays on terminal the result of strong junction tree computation for test purpose only.

eraseAllEvidence()

template<typename GUM_SCALAR >

virtual void gum::InfluenceDiagramInference< GUM_SCALAR >::eraseAllEvidence ( )

virtual

See also

gum::IInfluenceDiagramInference::eraseAllEvidence().

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

eraseEvidence()

template<typename GUM_SCALAR >

virtual void gum::InfluenceDiagramInference< GUM_SCALAR >::eraseEvidence ( const Potential< GUM_SCALAR > *  evidence )

virtual

See also

gum::IInfluenceDiagramInference::eraseEvidence().

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

getBestDecisionChoice()

template<typename GUM_SCALAR >

Idx gum::InfluenceDiagramInference< GUM_SCALAR >::getBestDecisionChoice ( NodeId  decisionId )

virtual

See also

gum::IInfluenceDiagramInference::getBestDecisionChoice( NodeId decisionId )

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

getMEU()

template<typename GUM_SCALAR >

GUM_SCALAR gum::InfluenceDiagramInference< GUM_SCALAR >::getMEU ( )

virtual

See also

gum::IInfluenceDiagramInference::getMEU().

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

getTriangulation()

template<typename GUM_SCALAR >

Triangulation& gum::InfluenceDiagramInference< GUM_SCALAR >::getTriangulation

(

)

Returns the Triangulation used by this class.

influenceDiagram()

template<typename GUM_SCALAR >

const InfluenceDiagram< GUM_SCALAR >& gum::IInfluenceDiagramInference< GUM_SCALAR >::influenceDiagram ( ) const

inherited

Returns a constant reference over the InfluenceDiagram on which this class work.

insertEvidence()

template<typename GUM_SCALAR >

virtual void gum::InfluenceDiagramInference< GUM_SCALAR >::insertEvidence ( const List< const Potential< GUM_SCALAR > * > &  evidenceList )

virtual

See also

gum::IInfluenceDiagramInference::insertEvidence().

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

makeInference()

template<typename GUM_SCALAR >

virtual void gum::InfluenceDiagramInference< GUM_SCALAR >::makeInference ( )

virtual

See also

gum::IInfluenceDiagramInference::makeInference().

Implements gum::IInfluenceDiagramInference< GUM_SCALAR >.

Member Data Documentation

__cliqueEliminationMap

template<typename GUM_SCALAR >

HashTable< Size, NodeId > gum::InfluenceDiagramInference< GUM_SCALAR >::__cliqueEliminationMap

private

Mapping of the nodes with the clique used to put their CPT.

Definition at line 164 of file influenceDiagramInference.h.

__cliquePropertiesMap

template<typename GUM_SCALAR >

NodeProperty< CliqueProperties< GUM_SCALAR >* > gum::InfluenceDiagramInference< GUM_SCALAR >::__cliquePropertiesMap

private

Mapping of the nodes with the clique used to put their CPT.

Definition at line 162 of file influenceDiagramInference.h.

__infDiag

template<typename GUM_SCALAR >

const InfluenceDiagram< GUM_SCALAR >& gum::IInfluenceDiagramInference< GUM_SCALAR >::__infDiag

protectedinherited

The Bayes net we wish to perform inference on.

Definition at line 111 of file IInfluenceDiagramInference.h.

__inferenceMade

template<typename GUM_SCALAR >

bool gum::InfluenceDiagramInference< GUM_SCALAR >::__inferenceMade

private

Mapping of the nodes with the clique used to put their CPT.

Definition at line 168 of file influenceDiagramInference.h.

__inferencePotential

template<typename GUM_SCALAR >

Potential< GUM_SCALAR >* gum::InfluenceDiagramInference< GUM_SCALAR >::__inferencePotential

private

The resulting potential from inference.

Definition at line 148 of file influenceDiagramInference.h.

__inferenceUtility

template<typename GUM_SCALAR >

Potential< GUM_SCALAR >* gum::InfluenceDiagramInference< GUM_SCALAR >::__inferenceUtility

private

The resulting utility from inference.

Definition at line 151 of file influenceDiagramInference.h.

__nodeToCliqueMap

template<typename GUM_SCALAR >

NodeProperty< NodeId > gum::InfluenceDiagramInference< GUM_SCALAR >::__nodeToCliqueMap

private

Mapping of the nodes with the clique used to put their CPT.

Definition at line 160 of file influenceDiagramInference.h.

__potentialDummies

template<typename GUM_SCALAR >

Set< Potential< GUM_SCALAR >* > gum::InfluenceDiagramInference< GUM_SCALAR >::__potentialDummies

private

The set of dummies sparse potential matrix created.

Definition at line 142 of file influenceDiagramInference.h.

__triangulation

template<typename GUM_SCALAR >

Triangulation* gum::InfluenceDiagramInference< GUM_SCALAR >::__triangulation

private

The triangulation algorithm.

Definition at line 139 of file influenceDiagramInference.h.

__utakenDecisionMap

template<typename GUM_SCALAR >

HashTable< NodeId, Idx > gum::InfluenceDiagramInference< GUM_SCALAR >::__utakenDecisionMap

private

Mapping of the nodes with the clique used to put their CPT.

Definition at line 166 of file influenceDiagramInference.h.

__utilityDummies

template<typename GUM_SCALAR >

Set< Potential< GUM_SCALAR >* > gum::InfluenceDiagramInference< GUM_SCALAR >::__utilityDummies

private

The set of dummies sparse utilities matrix created.

Definition at line 145 of file influenceDiagramInference.h.

---

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

• agrum/ID/inference/influenceDiagramInference.h