Package opennlp.tools.ml.maxent.quasinewton

Class QNTrainer

java.lang.Object

opennlp.tools.ml.AbstractTrainer

opennlp.tools.ml.AbstractEventTrainer

opennlp.tools.ml.maxent.quasinewton.QNTrainer

All Implemented Interfaces:

Trainer, EventTrainer

public class QNTrainer
extends AbstractEventTrainer

A Maxent model trainer using the L-BFGS algorithm.

See Also:

• AbstractEventTrainer
• QNMinimizer
• QNModel
• Trainer

Field Summary

Fields

Modifier and Type	Field	Description
static final double	L1COST_DEFAULT	The default L1-cost value is 0.1d.
static final String	L1COST_PARAM
static final double	L2COST_DEFAULT	The default L2-cost value is 0.1d.
static final String	L2COST_PARAM
static final int	M_DEFAULT	The default number of Hessian updates to store is 15.
static final String	M_PARAM
static final int	MAX_FCT_EVAL_DEFAULT	The default maximum number of function evaluations is 30,000.
static final String	MAX_FCT_EVAL_PARAM
static final String	MAXENT_QN_VALUE
static final int	THREADS_DEFAULT
static final String	THREADS_PARAM

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 interface opennlp.tools.ml.EventTrainer

EVENT_VALUE

Constructor Summary

Constructors

Constructor	Description
QNTrainer()	Initializes a QNTrainer.
QNTrainer(int m)	Initializes a QNTrainer with the specified parameter m.
QNTrainer(int m, int maxFctEval)	Initializes a QNTrainer with the specified parameters.
QNTrainer(TrainingParameters parameters)	Initializes a QNTrainer with the specified parameters.

Method Summary

Modifier and Type	Method	Description
AbstractModel	doTrain(DataIndexer indexer)
void	init(TrainingParameters trainingParameters, Map<String,String> reportMap)	Initializes a AbstractTrainer via TrainingParameters and a report map.
boolean	isSortAndMerge()
QNModel	trainModel(int iterations, DataIndexer indexer)	Trains a model using the QN algorithm.
void	validate()	Checks the configured parameters.

Methods inherited from class opennlp.tools.ml.AbstractEventTrainer

getDataIndexer, train, train

Methods inherited from class opennlp.tools.ml.AbstractTrainer

getAlgorithm, getCutoff, getIterations

Methods inherited from class java.lang.Object

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

Field Details

MAXENT_QN_VALUE

public static final String MAXENT_QN_VALUE

See Also:

• Constant Field Values

THREADS_PARAM

public static final String THREADS_PARAM

See Also:

• Constant Field Values

THREADS_DEFAULT

public static final int THREADS_DEFAULT

See Also:

• Constant Field Values

L1COST_PARAM

public static final String L1COST_PARAM

See Also:

• Constant Field Values

L1COST_DEFAULT

public static final double L1COST_DEFAULT

The default L1-cost value is 0.1d.

See Also:

• Constant Field Values

L2COST_PARAM

public static final String L2COST_PARAM

See Also:

• Constant Field Values

L2COST_DEFAULT

public static final double L2COST_DEFAULT

The default L2-cost value is 0.1d.

See Also:

• Constant Field Values

M_PARAM

public static final String M_PARAM

See Also:

• Constant Field Values

M_DEFAULT

public static final int M_DEFAULT

The default number of Hessian updates to store is 15.

See Also:

• Constant Field Values

MAX_FCT_EVAL_PARAM

public static final String MAX_FCT_EVAL_PARAM

See Also:

• Constant Field Values

MAX_FCT_EVAL_DEFAULT

public static final int MAX_FCT_EVAL_DEFAULT

The default maximum number of function evaluations is 30,000.

See Also:

• Constant Field Values

Constructor Details

QNTrainer

public QNTrainer()

Initializes a QNTrainer.

QNTrainer

public QNTrainer(TrainingParameters parameters)

Initializes a QNTrainer with the specified parameters.

Parameters:

parameters - The TrainingParameters to use.

QNTrainer

public QNTrainer(int m)

Initializes a QNTrainer with the specified parameter m.

Parameters:

m - The number of hessian updates to store.

QNTrainer

public QNTrainer(int m,
 int maxFctEval)

Initializes a QNTrainer with the specified parameters.

Parameters:

m - The number of hessian updates to store.

Method Details

init

public void init(TrainingParameters trainingParameters,
 Map<String,String> reportMap)

Description copied from class: AbstractTrainer

Initializes a AbstractTrainer via TrainingParameters and a report map.

Specified by:

init in interface Trainer

Overrides:

init in class AbstractTrainer

Parameters:

trainingParameters - The TrainingParameters to use.

reportMap - The Map instance used as report map.

validate

public void validate()

Description copied from class: AbstractTrainer

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

Overrides:

validate in class AbstractEventTrainer

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

trainModel

public QNModel trainModel(int iterations,
 DataIndexer indexer)

Trains a model using the QN algorithm.

Parameters:

iterations - The number of QN iterations to perform.

indexer - The DataIndexer used to compress events in memory.

Returns:

A trained QNModel which can be used immediately or saved to disk using an QNModelWriter.

Throws:

IllegalArgumentException - Thrown if parameters were invalid.