edu.ucla.belief.learn
Class Learning

Object
  extended by Learning

public class Learning
extends Object

A collection of static functions for performing learning on BeliefNetworks. This class cannot be instantiated.


Field Summary
static double DOUBLE_BIAS_VALUE
           
static String STR_EM_FILENAME_PREFIX
           
 
Constructor Summary
Learning()
           
 
Method Summary
static double getLastLikelihood()
          Returns the -log(Likelihood) for the last call to learnParamsEM().
static BeliefNetwork learnParamsEM(BeliefNetwork bn, LearningData data2, double threshold, int maxIterations, Dynamator dyn, boolean withBias)
           
static BeliefNetwork learnParamsEM(BeliefNetwork bn, LearningData data, Dynamator dynamator, boolean withBias)
          Returns a new BeliefNetwork with CPTs learned from one interation of the EM learning algorithm.
static Thread learnParamsEM(BeliefNetwork bn, LearningData data2, Dynamator dynamator, double threshold, int maxIterations, boolean withBias, ProgressMonitor pm, EMThread.LearningListener ll)
           
static BeliefNetwork learnParamsEM(BeliefNetwork bn, LearningDataOld data, double threshold, int maxIterations, Dynamator dyn, boolean withBias)
          Returns a new BeliefNetwork with CPTs learned from the EM learning algorithm.
static BeliefNetwork learnParamsEM(BeliefNetwork bn, LearningDataOld data, Dynamator dynamator, boolean withBias)
          Returns a new BeliefNetwork with CPTs learned from one interation of the EM learning algorithm.
static Thread learnParamsEM(BeliefNetwork bn, LearningDataOld data, Dynamator dynamator, double threshold, int maxIterations, boolean withBias, ProgressMonitor pm, EMThread.LearningListener ll)
           
static String renamePathForEmOutput(String oldHuginName)
           
 
Methods inherited from class Object
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

DOUBLE_BIAS_VALUE

public static final double DOUBLE_BIAS_VALUE
See Also:
Constant Field Values

STR_EM_FILENAME_PREFIX

public static final String STR_EM_FILENAME_PREFIX
See Also:
Constant Field Values
Constructor Detail

Learning

public Learning()
Method Detail

learnParamsEM

public static BeliefNetwork learnParamsEM(BeliefNetwork bn,
                                          LearningDataOld data,
                                          double threshold,
                                          int maxIterations,
                                          Dynamator dyn,
                                          boolean withBias)
                                   throws ArithmeticException
Returns a new BeliefNetwork with CPTs learned from the EM learning algorithm. Learning continues until the relative difference in the log-likelihoods from one iteration to the next are less than "threshold" or "maxIterations" is reached.

Parameters:
bn - Initial BeliefNetwork.
data - Complete or incomplete LearningDataOld for variables in "bn".
threshold - Convergence threshold.
maxIterations - Learning will repeat no more than this number of times.
Throws:
ArithmeticException

learnParamsEM

public static BeliefNetwork learnParamsEM(BeliefNetwork bn,
                                          LearningData data2,
                                          double threshold,
                                          int maxIterations,
                                          Dynamator dyn,
                                          boolean withBias)
                                   throws ArithmeticException
Throws:
ArithmeticException

learnParamsEM

public static Thread learnParamsEM(BeliefNetwork bn,
                                   LearningDataOld data,
                                   Dynamator dynamator,
                                   double threshold,
                                   int maxIterations,
                                   boolean withBias,
                                   ProgressMonitor pm,
                                   EMThread.LearningListener ll)
Parameters:
bn - Initial BeliefNetwork.
data - Complete or incomplete LearningDataOld for variables in "bn".
threshold - Convergence threshold.
maxIterations - Learning will repeat no more than this number of times.
pm - The progress monitor to update.
ll - The LearningListener that will be notified when the EM algorithm terminates.
Since:
052002 Creates a new BeliefNetwork with CPTs learned from the EM learning algorithm. Learning continues until the relative difference in the log-likelihoods from one iteration to the next are less than "threshold" or "maxIterations" is reached.

learnParamsEM

public static Thread learnParamsEM(BeliefNetwork bn,
                                   LearningData data2,
                                   Dynamator dynamator,
                                   double threshold,
                                   int maxIterations,
                                   boolean withBias,
                                   ProgressMonitor pm,
                                   EMThread.LearningListener ll)
Since:
20031105

learnParamsEM

public static BeliefNetwork learnParamsEM(BeliefNetwork bn,
                                          LearningDataOld data,
                                          Dynamator dynamator,
                                          boolean withBias)
                                   throws ArithmeticException
Returns a new BeliefNetwork with CPTs learned from one interation of the EM learning algorithm.

Parameters:
bn - Initial BeliefNetwork.
data - Complete or incomplete LearningDataOld for variables in "bn".
Throws:
ArithmeticException

learnParamsEM

public static BeliefNetwork learnParamsEM(BeliefNetwork bn,
                                          LearningData data,
                                          Dynamator dynamator,
                                          boolean withBias)
                                   throws ArithmeticException
Returns a new BeliefNetwork with CPTs learned from one interation of the EM learning algorithm.

Parameters:
bn - Initial BeliefNetwork.
data - Complete or incomplete LearningData for variables in "bn".
Throws:
ArithmeticException

getLastLikelihood

public static double getLastLikelihood()
Returns the -log(Likelihood) for the last call to learnParamsEM().


renamePathForEmOutput

public static String renamePathForEmOutput(String oldHuginName)
Since:
060105


Copyright 2010 UCLA Automated Reasoning Group