Package opennlp.tools.ml.perceptron

Class SimplePerceptronSequenceTrainer

java.lang.Object

opennlp.tools.ml.AbstractTrainer

opennlp.tools.ml.AbstractEventModelSequenceTrainer

opennlp.tools.ml.perceptron.SimplePerceptronSequenceTrainer

All Implemented Interfaces:

Trainer, EventModelSequenceTrainer<Event>

public class SimplePerceptronSequenceTrainer
extends AbstractEventModelSequenceTrainer

Trains models with sequences using the perceptron algorithm.

Each outcome is represented as a binary perceptron classifier. This supports standard (integer) weighting as well average weighting.

Sequence information is used in a simplified was to that described in: Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with the Perceptron Algorithm. Michael Collins, EMNLP 2002.

Specifically only updates are applied to tokens which were incorrectly tagged by a sequence tagger rather than to all feature across the sequence which differ from the training sequence.

See Also:

• PerceptronModel
• AbstractEventModelSequenceTrainer

Field Summary

Fields

Modifier and Type	Field	Description
static final String	PERCEPTRON_SEQUENCE_VALUE

Fields inherited from class opennlp.tools.ml.AbstractTrainer

ALGORITHM_PARAM, CUTOFF_DEFAULT, CUTOFF_PARAM, ITERATIONS_DEFAULT, ITERATIONS_PARAM, TRAINER_TYPE_PARAM

Fields inherited from interface opennlp.tools.ml.EventModelSequenceTrainer

SEQUENCE_VALUE

Constructor Summary

Constructors

Constructor	Description
SimplePerceptronSequenceTrainer()	Instantiates a SimplePerceptronSequenceTrainer with a default configuration of training parameters.

Method Summary

Modifier and Type	Method	Description
AbstractModel	doTrain(SequenceStream<Event> events)
void	nextIteration(int iteration)
AbstractModel	trainModel(int iterations, SequenceStream<Event> sequenceStream, int cutoff, boolean useAverage)	Trains a PerceptronModel with given parameters.
void	validate()	Checks the configured parameters.

Methods inherited from class opennlp.tools.ml.AbstractEventModelSequenceTrainer

train

Methods inherited from class opennlp.tools.ml.AbstractTrainer

getAlgorithm, getCutoff, getIterations, init

Methods inherited from class java.lang.Object

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

Methods inherited from interface opennlp.tools.commons.Trainer

init

Field Details

PERCEPTRON_SEQUENCE_VALUE

public static final String PERCEPTRON_SEQUENCE_VALUE

See Also:

• Constant Field Values

Constructor Details

SimplePerceptronSequenceTrainer

public SimplePerceptronSequenceTrainer()

Instantiates a SimplePerceptronSequenceTrainer with a default configuration of training parameters.

Method Details

validate

public void validate()

Checks the configured parameters. If a subclass overrides this, it should call super.validate();.

Overrides:

validate in class AbstractTrainer

Throws:

IllegalArgumentException - Thrown if the algorithm name is not equal to {PERCEPTRON_SEQUENCE_VALUE}.

doTrain

public AbstractModel doTrain(SequenceStream<Event> events)
                      throws IOException

Specified by:

doTrain in class AbstractEventModelSequenceTrainer

Throws:

IOException

trainModel

public AbstractModel trainModel(int iterations,
 SequenceStream<Event> sequenceStream,
 int cutoff,
 boolean useAverage)
                         throws IOException

Trains a PerceptronModel with given parameters.

Parameters:

iterations - The number of iterations to use for training.

sequenceStream - The SequenceStream<Event> used as data input.

cutoff - The {AbstractTrainer.CUTOFF_PARAM} value to use for training.

useAverage - Whether to use 'averaging', or not.

Returns:

A valid, trained perceptron model.

Throws:

IOException

nextIteration

public void nextIteration(int iteration)
                   throws IOException

Throws:

IOException