com.kpro.algorithm
Class LearnAlgorithm

java.lang.Object
  extended by com.kpro.algorithm.LearnAlgorithm
Direct Known Subclasses:
Learn_Constant, LearnAlgBasic, LearnAlgStand

public abstract class LearnAlgorithm
extends java.lang.Object

An abstract class covering all learning algorithms. The learning algorithm alters the weights configuration after examining the current database after the addition of a new policy.

Version:
29.09.11
Author:
ngerstle

Field Summary
(package private)  java.lang.String[] extraArgs
           
(package private)  java.util.Properties weightsConfig
           
 
Constructor Summary
LearnAlgorithm(java.util.Properties weightsConfig, java.lang.String[] extraArgs)
          Constructor for a learning algorithm.
 
Method Summary
protected abstract  java.util.Properties applyML(Gio theIO)
          The place to implement the actual algorithm for learning.
 void learn(Gio theIO)
          runs the learning algorithm, and puts the results in the newWeight buffer in theIO(Gio)
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

weightsConfig

java.util.Properties weightsConfig

extraArgs

java.lang.String[] extraArgs
Constructor Detail

LearnAlgorithm

public LearnAlgorithm(java.util.Properties weightsConfig,
                      java.lang.String[] extraArgs)
Constructor for a learning algorithm. accepts a weights configuration file.

Parameters:
weightsConfig - the weights configuration file
extraArgs - extra arguments defined in configuration file
Method Detail

learn

public void learn(Gio theIO)
runs the learning algorithm, and puts the results in the newWeight buffer in theIO(Gio)

Parameters:
theIO -

applyML

protected abstract java.util.Properties applyML(Gio theIO)
The place to implement the actual algorithm for learning.

Parameters:
theIO - the current database, via Gio
Returns:
a modified properties, such that it more accurately reflects the relationships between policies