edu.ksu.cis.bnj.bbn.learning.score
Class ShieldScore

java.lang.Object
  edu.ksu.cis.bnj.bbn.learning.LearnerScore
      edu.ksu.cis.bnj.bbn.learning.score.SparseScore
          edu.ksu.cis.bnj.bbn.learning.score.ShieldScore

public class ShieldScore

extends SparseScore

Measure_Shield score from Friedman, Nachman, and Pe'er [1999] Measure_Shield(X_i, X_j | B) = I(X_i ; X_j, Pa(X_i))

Author:

Roby Joehanes Last modified 05 Apr 2003

Field Summary

Fields inherited from class edu.ksu.cis.bnj.bbn.learning.score.SparseScore

arityCache, attributeSize, maxArity, nodes, totalRows, tupleSize

Fields inherited from class edu.ksu.cis.bnj.bbn.learning.LearnerScore

NO_CANDIDATES, owner, tally

Constructor Summary

ShieldScore()
default constructor for ShieldScore

ShieldScore(Learner l)
constructor for ShieldScore given a Learner object

Method Summary

double	getScore(int X_i, int X_j, java.util.Set[] parentTable) calculate mutual information I(X_i ; X_j, Pa(X_i)) = D_KL(\hat P(X_i, X_j, Pa(X_i)) || \hat P(X_i) * \hat P(X_j, Pa(X_i))) = \Sum_{X_i, X_j, Pa(X_i)} \hat P(X_i, X_j, Pa(X_i)) log \frac{\hat P(X_i, X_j, Pa(X_i))}{\hat P(X_i) * \hat P(X_j, Pa(X_i))}
void	initialize() This is just to reset the cache.
protected int	iterateUniqueInstantiation(int depth, Tally tally, int total) Recursively filter out the data according to the current node

Methods inherited from class edu.ksu.cis.bnj.bbn.learning.score.SparseScore

initializeCache

Methods inherited from class edu.ksu.cis.bnj.bbn.learning.LearnerScore

getTallyer, setData, setOwner

Methods inherited from class java.lang.Object

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

Constructor Detail

ShieldScore

public ShieldScore()

default constructor for ShieldScore

ShieldScore

public ShieldScore(Learner l)

constructor for ShieldScore given a Learner object

Parameters:

l -

Method Detail

getScore

public double getScore(int X_i,
                       int X_j,
                       java.util.Set[] parentTable)

calculate mutual information I(X_i ; X_j, Pa(X_i)) = D_KL(\hat P(X_i, X_j, Pa(X_i)) || \hat P(X_i) * \hat P(X_j, Pa(X_i))) = \Sum_{X_i, X_j, Pa(X_i)} \hat P(X_i, X_j, Pa(X_i)) log \frac{\hat P(X_i, X_j, Pa(X_i))}{\hat P(X_i) * \hat P(X_j, Pa(X_i))}

Specified by:

getScore in class LearnerScore

Parameters:

X_i - The index of the currently evaluated node

X_j - The index of the parent candidate

parentTable - The array of Sets of integers. Each integer in the set is the index of currently assembled parent of each node.

Returns:

double the score

See Also:

edu.ksu.cis.bnj.bbn.learning.LearnerScore#getScore(java.lang.String, java.util.Set)

iterateUniqueInstantiation

protected int iterateUniqueInstantiation(int depth,
                                         Tally tally,
                                         int total)

Recursively filter out the data according to the current node

Parameters:

tally -

initialize

public void initialize()

This is just to reset the cache. Since it is not dependent on tuple numbers, we can simply ignore this.

Overrides:

initialize in class LearnerScore

See Also:

LearnerScore.initialize()