Module es.upm.fi.cig.multictbnc

Package es.upm.fi.cig.multictbnc.models

Class CTBN<NodeType extends Node>

java.lang.Object

es.upm.fi.cig.multictbnc.models.AbstractPGM<NodeType>

es.upm.fi.cig.multictbnc.models.CTBN<NodeType>

Type Parameters:

NodeType - type of the nodes of the CTBN (e.g. nodes that learn a CIM)

All Implemented Interfaces:

PGM<NodeType>

public class CTBN<NodeType extends Node>
extends AbstractPGM<NodeType>

Implements a continuous-time Bayesian network (CTBN).

Constructor Summary

Constructors

Constructor	Description
CTBN(Dataset dataset, List<String> nameVariables, CTBNLearningAlgorithms ctbnLearningAlgs, StructureConstraints structureConstraints, BN<? extends Node> bnClassSubgraph, Class<NodeType> nodeClass)	Initialises a continuous-time Bayesian network given a dataset, the list of variables to use and the algorithms for parameter and structure learning.
CTBN(Dataset dataset, List<String> nameVariables, CTBNLearningAlgorithms ctbnLearningAlgs, StructureConstraints structureConstraints, Class<NodeType> nodeClass)	Initialises a continuous-time Bayesian network given a dataset, the list of variables to use and the algorithms for parameter and structure learning.
CTBN(CTBN<NodeType> ctbn, boolean cloneStructure)	Constructor to clone a continuous-time Bayesian network.
CTBN(CTBN<NodeType> ctbn, Dataset dataset)	Constructor to clone a continuous-time Bayesian network.
CTBN(List<NodeType> nodes, BN<? extends Node> bnClassSubgraph)	Initialises a continuous-time Bayesian network by receiving a list of nodes and a Bayesian network modelling the class subgraph of a Multi-CTBNC.
CTBN(List<NodeType> nodes, BN<? extends Node> bnClassSubgraph, Dataset dataset)	Initialises a continuous-time Bayesian network by receiving a list of nodes, a Bayesian network modelling the class subgraph of a Multi-CTBNC and a training dataset.

Method Summary

Modifier and Type	Method	Description
BN<? extends Node>	getBnClassSubgraph()	Returns the class subgraph (Bayesian network).
String	getModelIdentifier()	Returns a String that identifies the model.
String	getType()	Provides the type of PGM.
void	learnParameters(Dataset dataset)	Learns the parameters of the PGM using the provided dataset.
void	setStructure(int nodeIndex, boolean[][] adjacencyMatrix)	Modifies the structure of the continuous-time Bayesian network by changing the parent set of a specified node and updates its parameters.
void	setStructure(List<Integer> nodeIndexes, boolean[][] adjacencyMatrix)	Modifies the structure of the continuous-time Bayesian network by changing the parent set of some specified nodes and updates their parameters.
String	toString()

Methods inherited from class es.upm.fi.cig.multictbnc.models.AbstractPGM

addNodes, areParametersEstimated, computeSufficientStatistics, display, display, display, getAdjacencyMatrix, getDataset, getHyperparameters, getIndexNodes, getIndexOfNode, getNamesNodesByIndex, getNameVariables, getNodeByIndex, getNodeByName, getNodeClass, getNodeFactory, getNodeIndexer, getNodes, getNumNodes, getParameterLearningAlg, getStructureConstraints, getStructureLearningAlg, initialiseModel, isStructureLegal, learn, learn, learn, learn, learnParameters, learnParameters, learnParameters, learnParameters, learnParameters, removeAllEdges, removeAllNodes, saveGraph, setDataset, setNameVariables, setParameterLearningAlgorithm, setStructure, setStructureConstraints, setStructureLearningAlgorithm, setStructureModifiedNodes

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Details

CTBN

public CTBN(List<NodeType> nodes,
 BN<? extends Node> bnClassSubgraph)

Initialises a continuous-time Bayesian network by receiving a list of nodes and a Bayesian network modelling the class subgraph of a Multi-CTBNC. This constructor is used when the structure of the model is provided and to build a MultiCTBNC.

Parameters:

nodes - nodes that make up the model and define its structure

bnClassSubgraph - a Bayesian network modelling the class subgraph of a Multi-CTBNC (necessary to estimate the conditional log-likelihood)

CTBN

public CTBN(List<NodeType> nodes,
 BN<? extends Node> bnClassSubgraph,
 Dataset dataset)

Initialises a continuous-time Bayesian network by receiving a list of nodes, a Bayesian network modelling the class subgraph of a Multi-CTBNC and a training dataset. This constructor is used when the structure of the model is provided and to build a MultiCTBNC.

Parameters:

nodes - nodes that make up the model and define its structure

bnClassSubgraph - a Bayesian network modelling the class subgraph of a Multi-CTBNC (necessary to estimate the conditional log-likelihood)

dataset - dataset used to learn the continuous-time Bayesian network

CTBN

public CTBN(Dataset dataset,
 List<String> nameVariables,
 CTBNLearningAlgorithms ctbnLearningAlgs,
 StructureConstraints structureConstraints,
 BN<? extends Node> bnClassSubgraph,
 Class<NodeType> nodeClass)

Initialises a continuous-time Bayesian network given a dataset, the list of variables to use and the algorithms for parameter and structure learning. This constructor was thought to be used by a MultiCTBNC.

Parameters:

dataset - dataset used to learn the continuous-time Bayesian network

nameVariables - variables that makeup the model

ctbnLearningAlgs - a CTBNLearningAlgorithms containing the algorithms for parameter and structure learning

structureConstraints - structure constraints to take into account during the learning of the continuous-time Bayesian network

bnClassSubgraph - a Bayesian network modelling the class subgraph of a Multi-CTBNC (necessary to estimate the conditional log-likelihood)

nodeClass - type of the CTBN nodes

CTBN

public CTBN(Dataset dataset,
 List<String> nameVariables,
 CTBNLearningAlgorithms ctbnLearningAlgs,
 StructureConstraints structureConstraints,
 Class<NodeType> nodeClass)

Initialises a continuous-time Bayesian network given a dataset, the list of variables to use and the algorithms for parameter and structure learning.

Parameters:

dataset - dataset used to learn the continuous-time Bayesian network

nameVariables - variables that makeup the model

ctbnLearningAlgs - a CTBNLearningAlgorithms containing the algorithms for parameter and structure learning

structureConstraints - structure constrains to take into account during the learning of the continuous-time Bayesian network

nodeClass - type of the CTBN nodes

CTBN

public CTBN(CTBN<NodeType> ctbn,
 boolean cloneStructure)

Constructor to clone a continuous-time Bayesian network.

Parameters:

ctbn - continuous-time Bayesian network to clone

cloneStructure - true if the structure of the CTBN should be cloned, false to define the model nodes from a dataset contained in the CTBN, if any, without copying the structure.

CTBN

public CTBN(CTBN<NodeType> ctbn,
 Dataset dataset)

Constructor to clone a continuous-time Bayesian network. It clones the nodes of the CTBN and save the provided dataset.

Parameters:

ctbn - continuous-time Bayesian network

dataset - a dataset

Method Details

getBnClassSubgraph

public BN<? extends Node> getBnClassSubgraph()

Returns the class subgraph (Bayesian network). This is necessary when the structure is defined by optimising the conditional log-likelihood.

Returns:

Bayesian network modelling the class subgraph

learnParameters

public void learnParameters(Dataset dataset)

Description copied from interface: PGM

Learns the parameters of the PGM using the provided dataset.

Specified by:

learnParameters in interface PGM<NodeType extends Node>

Overrides:

learnParameters in class AbstractPGM<NodeType extends Node>

Parameters:

dataset - dataset used to learn the parameters

setStructure

public void setStructure(int nodeIndex,
 boolean[][] adjacencyMatrix)

Modifies the structure of the continuous-time Bayesian network by changing the parent set of a specified node and updates its parameters. This method is necessary to learn the model structure by optimising the parent set of its nodes.

Specified by:

setStructure in interface PGM<NodeType extends Node>

Overrides:

setStructure in class AbstractPGM<NodeType extends Node>

Parameters:

nodeIndex - node index whose parent set is modified

adjacencyMatrix - adjacency matrix of the model containing the new parent set of the specified node

setStructure

public void setStructure(List<Integer> nodeIndexes,
 boolean[][] adjacencyMatrix)

Modifies the structure of the continuous-time Bayesian network by changing the parent set of some specified nodes and updates their parameters.

Specified by:

setStructure in interface PGM<NodeType extends Node>

Overrides:

setStructure in class AbstractPGM<NodeType extends Node>

Parameters:

nodeIndexes - node indexes whose parent sets are modified

adjacencyMatrix - adjacency matrix of the model containing the new parent set of the specified nodes

getModelIdentifier

public String getModelIdentifier()

Description copied from interface: PGM

Returns a String that identifies the model.

Returns:

String that identifies the model

getType

public String getType()

Description copied from interface: PGM

Provides the type of PGM.

Returns:

string describing the type of PGM

toString

public String toString()

Overrides:

toString in class Object