trainableSegmentation.metrics

Class InformationStatistics

java.lang.Object

trainableSegmentation.metrics.InformationStatistics

public class InformationStatistics
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
double	fScore F-score, harmonic mean of precision and recall
double	hs entropy of predicted segmentation S
double	hst conditional entropy of S given T
double	ht entropy of ground truth T
double	hts conditional entropy of T given S
double	precision precision: C(T|S) = ( H(T) - H(T|S) ) / H(T)
double	recall recall C(S|T) = ( H(S) - H(S|T) ) / H(S)
double	vi value of the variation of information

Constructor Summary

Constructors

Constructor and Description
InformationStatistics(double ht, double hs, double hts, double hst, double vi) Create information statistics

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

ht

public double ht

entropy of ground truth T

hs

public double hs

entropy of predicted segmentation S

hts

public double hts

conditional entropy of T given S

hst

public double hst

conditional entropy of S given T

precision

public double precision

precision: C(T|S) = ( H(T) - H(T|S) ) / H(T)

recall

public double recall

recall C(S|T) = ( H(S) - H(S|T) ) / H(S)

fScore

public double fScore

F-score, harmonic mean of precision and recall

vi

public double vi

value of the variation of information

Constructor Detail

InformationStatistics

public InformationStatistics(double ht,
                             double hs,
                             double hts,
                             double hst,
                             double vi)

Create information statistics

Parameters:

ht - entropy of ground truth T

hs - entropy of predicted segmentation S

hts - conditional entropy of T given S

hst - conditional entropy of S given T

vi - value of the variation of information