dragon.ml.seqmodel.crf

Class LBFGSBasicTrainer

java.lang.Object

dragon.ml.seqmodel.crf.AbstractCRF

dragon.ml.seqmodel.crf.AbstractTrainer

dragon.ml.seqmodel.crf.LBFGSBasicTrainer

All Implemented Interfaces:

Trainer

Direct Known Subclasses:

LBFGSSegmentTrainer

public class LBFGSBasicTrainer
extends AbstractTrainer

LBFGS basic trainer

Copyright: Copyright (c) 2005

Company: IST, Drexel University

Version:

1.0

Author:

Davis Zhou

Field Summary

Fields

Modifier and Type	Field and Description
protected double	epsForConvergence
protected double	invSigmaSquare
protected int	mForHessian

Fields inherited from class dragon.ml.seqmodel.crf.AbstractTrainer

doScaling, maxIteration, xtol

Fields inherited from class dragon.ml.seqmodel.crf.AbstractCRF

edgeGen, featureGenerator, lambda, model

Constructor Summary

Constructors

Constructor and Description
LBFGSBasicTrainer(ModelGraph model, FeatureGenerator featureGenerator)

Method Summary

Modifier and Type	Method and Description
protected double	computeFunctionGradient(Dataset diter, double[] lambda, double[] grad)
protected double	norm(double[] ar)
void	setAccuracy(int eps)
void	setGradientHistory(int history)
void	setInvSigmaSquare(int invSigmaSquare)
boolean	train(Dataset dataset) Trains the CRF model with labeled dataset

Methods inherited from class dragon.ml.seqmodel.crf.AbstractTrainer

genStateVector, genStateVectorLog, getMaxIteration, needScaling, setMaxIteration, setScalingOption

Methods inherited from class dragon.ml.seqmodel.crf.AbstractCRF

computeTransMatrix, computeTransMatrix, getFeatureGenerator, getModelGraph, getModelParameter, readModelParameter, saveModelParameter

Methods inherited from class java.lang.Object

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

Methods inherited from interface dragon.ml.seqmodel.crf.Trainer

getFeatureGenerator, getModelGraph, getModelParameter, saveModelParameter

Field Detail

mForHessian

protected int mForHessian

epsForConvergence

protected double epsForConvergence

invSigmaSquare

protected double invSigmaSquare

Constructor Detail

LBFGSBasicTrainer

public LBFGSBasicTrainer(ModelGraph model,
                         FeatureGenerator featureGenerator)

Method Detail

setGradientHistory

public void setGradientHistory(int history)

setAccuracy

public void setAccuracy(int eps)

setInvSigmaSquare

public void setInvSigmaSquare(int invSigmaSquare)

train

public boolean train(Dataset dataset)

Description copied from interface: Trainer

Trains the CRF model with labeled dataset

Parameters:

dataset - the dataset for training

Returns:

true if trained successfully

norm

protected double norm(double[] ar)

computeFunctionGradient

protected double computeFunctionGradient(Dataset diter,
                                         double[] lambda,
                                         double[] grad)