opennlp.tools.ml.perceptron

Class PerceptronTrainer

java.lang.Object

opennlp.tools.ml.AbstractTrainer

opennlp.tools.ml.AbstractEventTrainer

opennlp.tools.ml.perceptron.PerceptronTrainer

All Implemented Interfaces:

EventTrainer

public class PerceptronTrainer
extends AbstractEventTrainer

Trains models using the perceptron algorithm. Each outcome is represented as a binary perceptron classifier. This supports standard (integer) weighting as well average weighting as described in: Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with the Perceptron Algorithm. Michael Collins, EMNLP 2002.

Field Summary

Fields

Modifier and Type	Field and Description
static String	PERCEPTRON_VALUE
static double	TOLERANCE_DEFAULT

Fields inherited from class opennlp.tools.ml.AbstractEventTrainer

DATA_INDEXER_ONE_PASS_REAL_VALUE, DATA_INDEXER_ONE_PASS_VALUE, DATA_INDEXER_PARAM, DATA_INDEXER_TWO_PASS_VALUE

Fields inherited from class opennlp.tools.ml.AbstractTrainer

ALGORITHM_PARAM, CUTOFF_DEFAULT, CUTOFF_PARAM, ITERATIONS_DEFAULT, ITERATIONS_PARAM, TRAINER_TYPE_PARAM, VERBOSE_DEFAULT, VERBOSE_PARAM

Fields inherited from interface opennlp.tools.ml.EventTrainer

EVENT_VALUE

Constructor Summary

Constructors

Constructor and Description
PerceptronTrainer()
PerceptronTrainer(TrainingParameters parameters)

Method Summary

Modifier and Type	Method and Description
AbstractModel	doTrain(DataIndexer indexer)
boolean	isSortAndMerge()
boolean	isValid() Deprecated.
void	setSkippedAveraging(boolean averaging) Enables skipped averaging, this flag changes the standard averaging to special averaging instead.
void	setStepSizeDecrease(double decrease) Enables and sets step size decrease.
void	setTolerance(double tolerance) Specifies the tolerance.
AbstractModel	trainModel(int iterations, DataIndexer di, int cutoff)
AbstractModel	trainModel(int iterations, DataIndexer di, int cutoff, boolean useAverage)
void	validate() Check parameters.

Methods inherited from class opennlp.tools.ml.AbstractEventTrainer

getDataIndexer, train, train

Methods inherited from class opennlp.tools.ml.AbstractTrainer

getAlgorithm, getCutoff, getIterations, init, init

Methods inherited from class java.lang.Object

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

Methods inherited from interface opennlp.tools.ml.EventTrainer

init, init

Field Detail

PERCEPTRON_VALUE

public static final String PERCEPTRON_VALUE

See Also:

Constant Field Values

TOLERANCE_DEFAULT

public static final double TOLERANCE_DEFAULT

See Also:

Constant Field Values

Constructor Detail

PerceptronTrainer

public PerceptronTrainer()

PerceptronTrainer

public PerceptronTrainer(TrainingParameters parameters)

Method Detail

validate

public void validate()

Description copied from class: AbstractTrainer

Check parameters. If subclass overrides this, it should call super.validate();

Overrides:

validate in class AbstractEventTrainer

isValid

@Deprecated
public boolean isValid()

Deprecated.

Overrides:

isValid in class AbstractEventTrainer

Returns:

isSortAndMerge

public boolean isSortAndMerge()

Specified by:

isSortAndMerge in class AbstractEventTrainer

doTrain

public AbstractModel doTrain(DataIndexer indexer)
                      throws IOException

Specified by:

doTrain in class AbstractEventTrainer

Throws:

IOException

setTolerance

public void setTolerance(double tolerance)

Specifies the tolerance. If the change in training set accuracy is less than this, stop iterating.

Parameters:

tolerance -

setStepSizeDecrease

public void setStepSizeDecrease(double decrease)

Enables and sets step size decrease. The step size is decreased every iteration by the specified value.

Parameters:

decrease - - step size decrease in percent

setSkippedAveraging

public void setSkippedAveraging(boolean averaging)

Enables skipped averaging, this flag changes the standard averaging to special averaging instead.

If we are doing averaging, and the current iteration is one of the first 20 or it is a perfect square, then updated the summed parameters.

The reason we don't take all of them is that the parameters change less toward the end of training, so they drown out the contributions of the more volatile early iterations. The use of perfect squares allows us to sample from successively farther apart iterations.

Parameters:

averaging - averaging flag

trainModel

public AbstractModel trainModel(int iterations,
                                DataIndexer di,
                                int cutoff)

trainModel

public AbstractModel trainModel(int iterations,
                                DataIndexer di,
                                int cutoff,
                                boolean useAverage)