com.ibm.bluej.consistency.learning

Class VariationRank

java.lang.Object

com.ibm.bluej.consistency.learning.VariationRank

public class VariationRank
extends java.lang.Object

Field Summary

Fields

Modifier and Type	Field and Description
boolean	errorAnalysis
LearningState	learningState
double	margin
static double	RECENCY
double	recencyWeightedAccuracy
double	recencyWeightedUpdateCount
int	vcount

Constructor Summary

Constructors

Constructor and Description
VariationRank(LearningState learningState)

Method Summary

Methods

Modifier and Type	Method and Description
void	finish()
void	multiVariationRank(java.util.ArrayList<WorldWeightParam> good, java.util.ArrayList<WorldWeightParam> bad) We want one of the good to be better than all of the bad (perhaps by a margin) If this is not so, we badDelta on the good/bad pairs
void	variationRank(java.util.ArrayList<com.ibm.bluej.util.common.Pair<WorldWeightParam,java.lang.Double>> variations)

Methods inherited from class java.lang.Object

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

Field Detail

learningState

public LearningState learningState

RECENCY

public static final double RECENCY

See Also:

Constant Field Values

recencyWeightedUpdateCount

public double recencyWeightedUpdateCount

recencyWeightedAccuracy

public double recencyWeightedAccuracy

vcount

public int vcount

margin

public double margin

errorAnalysis

public boolean errorAnalysis

Constructor Detail

VariationRank

public VariationRank(LearningState learningState)

Method Detail

multiVariationRank

public void multiVariationRank(java.util.ArrayList<WorldWeightParam> good,
                      java.util.ArrayList<WorldWeightParam> bad)

We want one of the good to be better than all of the bad (perhaps by a margin) If this is not so, we badDelta on the good/bad pairs

Parameters:

good -

bad -

variationRank

public void variationRank(java.util.ArrayList<com.ibm.bluej.util.common.Pair<WorldWeightParam,java.lang.Double>> variations)

finish

public void finish()