cc.mallet.fst
Class CRFTrainerByL1LabelLikelihood

java.lang.Object
  cc.mallet.fst.TransducerTrainer
      cc.mallet.fst.CRFTrainerByLabelLikelihood
          cc.mallet.fst.CRFTrainerByL1LabelLikelihood

All Implemented Interfaces:

TransducerTrainer.ByOptimization

public class CRFTrainerByL1LabelLikelihood
extends CRFTrainerByLabelLikelihood

CRF trainer that implements L1-regularization.

Author:

Kedar Bellare

Nested Class Summary

Nested classes/interfaces inherited from class cc.mallet.fst.TransducerTrainer

TransducerTrainer.ByIncrements, TransducerTrainer.ByInstanceIncrements, TransducerTrainer.ByOptimization

Field Summary

Fields inherited from class cc.mallet.fst.CRFTrainerByLabelLikelihood

printGradient

Constructor Summary

CRFTrainerByL1LabelLikelihood(CRF crf)

CRFTrainerByL1LabelLikelihood(CRF crf, double l1Weight)
Constructor for CRF trainer.

Method Summary

Optimizer	getOptimizer(InstanceList trainingSet)
void	setL1RegularizationWeight(double l1Weight)

Methods inherited from class cc.mallet.fst.CRFTrainerByLabelLikelihood

getCRF, getGaussianPriorVariance, getIteration, getOptimizableCRF, getOptimizer, getTransducer, getUseHyperbolicPriorSharpness, getUseHyperbolicPriorSlope, getUseSparseWeights, isConverged, isFinishedTraining, setAddNoFactors, setGaussianPriorVariance, setHyperbolicPriorSharpness, setHyperbolicPriorSlope, setUseHyperbolicPrior, setUseSomeUnsupportedTrick, setUseSparseWeights, train, train, trainIncremental, trainWithFeatureInduction, trainWithFeatureInduction

Methods inherited from class cc.mallet.fst.TransducerTrainer

addEvaluator, addEvaluators, removeEvaluator, runEvaluators, train

Methods inherited from class java.lang.Object

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

Constructor Detail

CRFTrainerByL1LabelLikelihood

public CRFTrainerByL1LabelLikelihood(CRF crf)

CRFTrainerByL1LabelLikelihood

public CRFTrainerByL1LabelLikelihood(CRF crf,
                                     double l1Weight)

Constructor for CRF trainer.

Parameters:

crf - CRF to train.

l1Weight - Weight of L1 term in objective (l1Weight*|w|). Higher L1 weight means sparser solutions.

Method Detail

setL1RegularizationWeight

public void setL1RegularizationWeight(double l1Weight)

getOptimizer

public Optimizer getOptimizer(InstanceList trainingSet)

Overrides:

getOptimizer in class CRFTrainerByLabelLikelihood