gov.sandia.cognition.learning.algorithm.confidence

Class AdaptiveRegularizationOfWeights

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.learning.algorithm.AbstractBatchAndIncrementalLearner<InputOutputPair<? extends InputType,OutputType>,ResultType>

gov.sandia.cognition.learning.algorithm.AbstractSupervisedBatchAndIncrementalLearner<Vectorizable,java.lang.Boolean,DefaultConfidenceWeightedBinaryCategorizer>

gov.sandia.cognition.learning.algorithm.confidence.AdaptiveRegularizationOfWeights

All Implemented Interfaces:

BatchAndIncrementalLearner<InputOutputPair<? extends Vectorizable,java.lang.Boolean>,DefaultConfidenceWeightedBinaryCategorizer>, BatchLearner<java.util.Collection<? extends InputOutputPair<? extends Vectorizable,java.lang.Boolean>>,DefaultConfidenceWeightedBinaryCategorizer>, IncrementalLearner<InputOutputPair<? extends Vectorizable,java.lang.Boolean>,DefaultConfidenceWeightedBinaryCategorizer>, SupervisedBatchAndIncrementalLearner<Vectorizable,java.lang.Boolean,DefaultConfidenceWeightedBinaryCategorizer>, SupervisedBatchLearner<Vectorizable,java.lang.Boolean,DefaultConfidenceWeightedBinaryCategorizer>, SupervisedIncrementalLearner<Vectorizable,java.lang.Boolean,DefaultConfidenceWeightedBinaryCategorizer>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

@PublicationReference(author={"Koby Crammer","Alex Kulesza","Mark Dredze"},
                      title="Adaptive Regularization of Weight Vectors",
                      year=2009,
                      type=Conference,
                      publication="Advances in Neural Information Processing Systems",
                      url="http://papers.nips.cc/paper/3848-adaptive-regularization-of-weight-vectors.pdf")
public class AdaptiveRegularizationOfWeights
extends AbstractSupervisedBatchAndIncrementalLearner<Vectorizable,java.lang.Boolean,DefaultConfidenceWeightedBinaryCategorizer>

An implementation of the Adaptive Regularization of Weights (AROW) algorithm for online learning of a linear binary categorizer. It is a confidence-weighted algorithm that keeps track of the full covariance matrix when updating the learner.

Since:

3.3.0

Author:

Justin Basilico

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static double	DEFAULT_R The default value of r is 0.001.
protected double	r The r parameter that controls regularization weight.

Constructor Summary

Constructors

Constructor and Description
AdaptiveRegularizationOfWeights() Creates a new AdaptiveRegularizationOfWeights with default parameters.
AdaptiveRegularizationOfWeights(double r) Creates a new AdaptiveRegularizationOfWeights with the given parameters

Method Summary

Modifier and Type	Method and Description
DefaultConfidenceWeightedBinaryCategorizer	createInitialLearnedObject() Creates a new initial learned object, before any data is given.
double	getR() Gets the regularization parameter.
void	setR(double r) Sets the regularization parameter.
void	update(DefaultConfidenceWeightedBinaryCategorizer target, Vector input, boolean label) Perform an update for the target using the given input and associated label.
void	update(DefaultConfidenceWeightedBinaryCategorizer target, Vectorizable input, java.lang.Boolean output) The update method updates an object of ResultType using the given a new supervised input-output pair, using some form of "learning" algorithm.

Methods inherited from class gov.sandia.cognition.learning.algorithm.AbstractSupervisedBatchAndIncrementalLearner

update

Methods inherited from class gov.sandia.cognition.learning.algorithm.AbstractBatchAndIncrementalLearner

clone, learn, learn, update

Methods inherited from class java.lang.Object

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

Methods inherited from interface gov.sandia.cognition.learning.algorithm.BatchAndIncrementalLearner

learn

Methods inherited from interface gov.sandia.cognition.learning.algorithm.BatchLearner

learn

Methods inherited from interface gov.sandia.cognition.learning.algorithm.IncrementalLearner

update

Methods inherited from interface gov.sandia.cognition.util.CloneableSerializable

clone

Field Detail

DEFAULT_R

public static final double DEFAULT_R

The default value of r is 0.001.

See Also:

Constant Field Values

r

protected double r

The r parameter that controls regularization weight. Must be positive.

Constructor Detail

AdaptiveRegularizationOfWeights

public AdaptiveRegularizationOfWeights()

Creates a new AdaptiveRegularizationOfWeights with default parameters.

AdaptiveRegularizationOfWeights

public AdaptiveRegularizationOfWeights(double r)

Creates a new AdaptiveRegularizationOfWeights with the given parameters

Parameters:

r - The regularization parameter. Must be positive.

Method Detail

createInitialLearnedObject

public DefaultConfidenceWeightedBinaryCategorizer createInitialLearnedObject()

Description copied from interface: IncrementalLearner

Creates a new initial learned object, before any data is given.

Returns:

The initial learned object.

update

public void update(DefaultConfidenceWeightedBinaryCategorizer target,
                   Vectorizable input,
                   java.lang.Boolean output)

Description copied from interface: SupervisedIncrementalLearner

The update method updates an object of ResultType using the given a new supervised input-output pair, using some form of "learning" algorithm.

Parameters:

target - The object to update.

input - The supervised input to learn from.

output - The supervised output to learn from.

update

public void update(DefaultConfidenceWeightedBinaryCategorizer target,
                   Vector input,
                   boolean label)

Perform an update for the target using the given input and associated label.

Parameters:

target - The target to update.

input - The input value.

label - The label associated with the input.

getR

public double getR()

Gets the regularization parameter.

Returns:

The regularization parameter. Must be positive.

setR

public void setR(double r)

Sets the regularization parameter.

Parameters:

r - The regularization parameter. Must be positive.