unbbayes.evaluation
Class FastLWApproximateEvaluation

java.lang.Object
  unbbayes.evaluation.AEvaluation
      unbbayes.evaluation.FastLWApproximateEvaluation

All Implemented Interfaces:

IEvaluation, ILongTaskProgressObservable, ILongTaskProgressObserver

public class FastLWApproximateEvaluation
extends AEvaluation
implements ILongTaskProgressObserver

Field Summary

protected ILikelihoodWeightingSampling	lw
protected int[]	positionEvidenceNodeList
protected List<Node>	positionNodeList
protected int[]	positionTargetNodeList
protected byte[][]	sampleMatrix
protected int	sampleSize
protected float[]	sampleWeight
protected TreeVariable	targetNode

Fields inherited from class unbbayes.evaluation.AEvaluation

currentProgress, currentProgressStatus, evidenceEvaluationList, evidenceNodeList, evidenceSetCM, evidenceSetPCC, evidenceStatesProduct, factors, lastProgressUpdated, maxProgress, net, statesProduct, targetNodeList, targetStatesProduct, UNSET_VALUE

Constructor Summary

FastLWApproximateEvaluation(int sampleSize)

Method Summary

protected float[][]	computeCM(List<String> targetNodeNameList, List<String> evidenceNodeNameList)
protected void	evaluate(List<String> targetNodeNameList, List<String> evidenceNodeNameList, boolean onlyGCM)
float	getError()
static void	main(String[] args)
protected void	setMaxProgress(int evidenceNodeListSize) Compute an approximation of the maximum size for the progress of this long task and set it as the value of maxProgress.
void	update(LongTaskProgressChangedEvent status) To have a better approximation of the current progress, we have to keep track of the sampling process used in this long task.

Methods inherited from class unbbayes.evaluation.AEvaluation

computeFactors, evaluate, evaluate, getBestIndividualCostRate, getBestIndividualPCC, getBestMarginalImprovement, getCurrentProgress, getCurrentProgressStatus, getEvidenceEvaluationList, getEvidenceLinearCoord, getEvidenceSetCM, getEvidenceSetPCC, getLinearCoord, getMaxProgress, getMultidimensionalCoord, getPercentageDone, init, loadNetwork, notityObservers, registerObserver, removeObserver, show, updateProgress, updateProgress

Methods inherited from class java.lang.Object

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

Field Detail

sampleSize

protected int sampleSize

lw

protected ILikelihoodWeightingSampling lw

targetNode

protected TreeVariable targetNode

sampleMatrix

protected byte[][] sampleMatrix

sampleWeight

protected float[] sampleWeight

positionNodeList

protected List<Node> positionNodeList

positionTargetNodeList

protected int[] positionTargetNodeList

positionEvidenceNodeList

protected int[] positionEvidenceNodeList

Constructor Detail

FastLWApproximateEvaluation

public FastLWApproximateEvaluation(int sampleSize)

Method Detail

getError

public float getError()

Specified by:

getError in interface IEvaluation

Specified by:

getError in class AEvaluation

evaluate

protected void evaluate(List<String> targetNodeNameList,
                        List<String> evidenceNodeNameList,
                        boolean onlyGCM)
                 throws EvaluationException

Overrides:

evaluate in class AEvaluation

Throws:

EvaluationException

computeCM

protected float[][] computeCM(List<String> targetNodeNameList,
                              List<String> evidenceNodeNameList)
                       throws EvaluationException

Specified by:

computeCM in class AEvaluation

Throws:

EvaluationException

setMaxProgress

protected void setMaxProgress(int evidenceNodeListSize)

Compute an approximation of the maximum size for the progress of this long task and set it as the value of maxProgress.

update

public void update(LongTaskProgressChangedEvent status)

To have a better approximation of the current progress, we have to keep track of the sampling process used in this long task.

Specified by:

update in interface ILongTaskProgressObserver

main

public static void main(String[] args)
                 throws Exception

Throws:

Exception