cc.mallet.grmm.learning
Class PwplACRFTrainer

java.lang.Object
  cc.mallet.grmm.learning.DefaultAcrfTrainer
      cc.mallet.grmm.learning.PwplACRFTrainer

All Implemented Interfaces:

ACRFTrainer

public class PwplACRFTrainer
extends DefaultAcrfTrainer

Implementation of piecewise PL (Sutton and McCallum, 2007) NB The wrong-wrong options are for an extension that we tried that never quite worked Created: Mar 15, 2005

Version:

$Id: BiconditionalPiecewiseACRFTrainer.java,v 1.1 2007/10/22 21:37:40 mccallum Exp $

Author:

Nested Class Summary

class

PwplACRFTrainer.Maxable

Nested classes/interfaces inherited from class cc.mallet.grmm.learning.DefaultAcrfTrainer

DefaultAcrfTrainer.FileEvaluator, DefaultAcrfTrainer.LogEvaluator, DefaultAcrfTrainer.TestResults

Field Summary

static int	CONDITION_WW
static int	NO_WRONG_WRONG
static boolean	printGradient

Constructor Summary

PwplACRFTrainer()

Method Summary

Optimizable.ByGradientValue	createOptimizable(ACRF acrf, InstanceList training)
double	getWrongWrongThreshold()
static void	reportTrainingLikelihood(ACRF acrf, InstanceList trainingList)
void	setWrongWrongIter(int wrongWrongIter)
void	setWrongWrongThreshold(double wrongWrongThreshold)
void	setWrongWrongType(int wrongWrongType)
boolean	train(ACRF acrf, InstanceList trainingList, InstanceList validationList, InstanceList testSet, ACRFEvaluator eval, int numIter, Optimizable.ByGradientValue macrf)

Methods inherited from class cc.mallet.grmm.learning.DefaultAcrfTrainer

callEvaluator, getMaxer, getRandom, incrementalTrain, incrementalTrain, isRethrowExceptions, setMaxer, setOutputPrefix, setRethrowExceptions, someUnsupportedTrain, test, test, train, train, train, train, train, train

Methods inherited from class java.lang.Object

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

Field Detail

printGradient

public static boolean printGradient

NO_WRONG_WRONG

public static final int NO_WRONG_WRONG

See Also:

Constant Field Values

CONDITION_WW

public static final int CONDITION_WW

See Also:

Constant Field Values

Constructor Detail

PwplACRFTrainer

public PwplACRFTrainer()

Method Detail

createOptimizable

public Optimizable.ByGradientValue createOptimizable(ACRF acrf,
                                                     InstanceList training)

Overrides:

createOptimizable in class DefaultAcrfTrainer

getWrongWrongThreshold

public double getWrongWrongThreshold()

setWrongWrongThreshold

public void setWrongWrongThreshold(double wrongWrongThreshold)

setWrongWrongType

public void setWrongWrongType(int wrongWrongType)

setWrongWrongIter

public void setWrongWrongIter(int wrongWrongIter)

train

public boolean train(ACRF acrf,
                     InstanceList trainingList,
                     InstanceList validationList,
                     InstanceList testSet,
                     ACRFEvaluator eval,
                     int numIter,
                     Optimizable.ByGradientValue macrf)

Specified by:

train in interface ACRFTrainer

Overrides:

train in class DefaultAcrfTrainer

reportTrainingLikelihood

public static void reportTrainingLikelihood(ACRF acrf,
                                            InstanceList trainingList)