com.bayesserver.learning.structure

Class SearchStructuralLearning

java.lang.Object

com.bayesserver.learning.structure.SearchStructuralLearning

All Implemented Interfaces:

StructuralLearning

public final class SearchStructuralLearning
extends Object
implements StructuralLearning

A structural learning algorithm for Bayesian networks based on Search and Score. This algorithm performs a search of possible structures based on moves that are evaluated using a scoring function.

Constructor Summary

Constructors

Constructor and Description
SearchStructuralLearning()

Method Summary

Modifier and Type	Method and Description
StructuralLearningOutput	learn(EvidenceReaderCommandFactory evidenceReaderCommandFactory, List<Node> nodes, StructuralLearningOptions options) Learn the structure (links) of a Bayesian network.
StructuralLearningOutput	learn(EvidenceReaderCommand readerCommand, List<Node> nodes, StructuralLearningOptions options) Learn the structure (links) of a Bayesian network.

Methods inherited from class java.lang.Object

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

Constructor Detail

SearchStructuralLearning

public SearchStructuralLearning()

Method Detail

learn

public StructuralLearningOutput learn(EvidenceReaderCommand readerCommand,
                                      List<Node> nodes,
                                      StructuralLearningOptions options)

Learn the structure (links) of a Bayesian network.

Specified by:

learn in interface StructuralLearning

Parameters:

readerCommand - Can create a reader which iterates round the data used to learn the network.

nodes - The nodes to be considered for links.

options - Options for the structural learning algorithm.

Returns:

The output generated by the structural learning algorithm.

learn

public StructuralLearningOutput learn(EvidenceReaderCommandFactory evidenceReaderCommandFactory,
                                      List<Node> nodes,
                                      StructuralLearningOptions options)

Learn the structure (links) of a Bayesian network.

Parameters:

evidenceReaderCommandFactory - Creates new evidence reader commands, which may be partitioned for use in cross validation.

nodes - The nodes to be considered for links.

options - Options for the structural learning algorithm.

Returns:

The output generated by the structural learning algorithm.