geneticWedge.gp
Class EvaluativeStoppingCriterion

java.lang.Object
  extended by geneticWedge.gp.StoppingCriterion
      extended by geneticWedge.gp.EvaluativeStoppingCriterion
Direct Known Subclasses:
FitnessStoppingCriterion, StagnationStoppingCriterion

public abstract class EvaluativeStoppingCriterion
extends StoppingCriterion

An EvaluativeStoppingCriterion checks some property of the fitness function to decide whether evolution should be stopped. This may be contrasted with FixedGenerationStoppingCriterion which does not implememt early stopping.


Constructor Summary
EvaluativeStoppingCriterion()
           
 
Method Summary
 void setMaximise(boolean maximiseValue)
          If true, it is assumed that a high-valued FitnessFunction is desirable.
 boolean usesValidation()
          Early Stopping may be carried out based on changes in the training fitness or the validation fitness.
 
Methods inherited from class geneticWedge.gp.StoppingCriterion
stopIfNoDiversity, stopIfNoDiversity, stopReached, toString
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Constructor Detail

EvaluativeStoppingCriterion

public EvaluativeStoppingCriterion()
Method Detail

setMaximise

public void setMaximise(boolean maximiseValue)
If true, it is assumed that a high-valued FitnessFunction is desirable. Therefore evolution would usually be stopped if the FitnessFunction is no longer increasing. On the other hand, if set to false evolution should be stopped when the FitnessFunction is no longer decreasing.


usesValidation

public boolean usesValidation()
Early Stopping may be carried out based on changes in the training fitness or the validation fitness. This function returns true if early stopping is assessed against validation data, false if assessed against training data.