cc.mallet.fst.semi_supervised
Class GEKLCriterion

java.lang.Object
  extended by cc.mallet.fst.semi_supervised.GECriterion
      extended by cc.mallet.fst.semi_supervised.GEKLCriterion

public class GEKLCriterion
extends GECriterion


Field Summary
 
Fields inherited from class cc.mallet.fst.semi_supervised.GECriterion
count, expectation, name, target, weight
 
Constructor Summary
GEKLCriterion(java.lang.String name, double[] target, double weight)
           
 
Method Summary
protected  double getGradientConstant(int labelIndex)
          Returns the target/expectation ratio required in lattice computations.
 
Methods inherited from class cc.mallet.fst.semi_supervised.GECriterion
getCount, getExpectation, getExpectation, getName, getTarget, getTarget, getWeight, setCount, setExpectation
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

GEKLCriterion

public GEKLCriterion(java.lang.String name,
                     double[] target,
                     double weight)
Method Detail

getGradientConstant

protected double getGradientConstant(int labelIndex)
Returns the target/expectation ratio required in lattice computations.

*Note*: The ratio is divided by the feature count if the label expectations have been normalized.

Specified by:
getGradientConstant in class GECriterion