moa.classifiers.meta

Class DACC

java.lang.Object

moa.AbstractMOAObject

moa.options.AbstractOptionHandler

moa.classifiers.AbstractClassifier

moa.classifiers.meta.DACC

All Implemented Interfaces:

Serializable, Classifier, AWTRenderable, MOAObject, OptionHandler

Direct Known Subclasses:

ADACC

public class DACC
extends AbstractClassifier

Dynamic Adaptation to Concept Changes. Ensemble method for data streams that adapts to concept changes. Reference: JABER, G., CORNUEJOLS, A., and TARROUX, P. A New On-Line Learning Method for Coping with Recurring Concepts: The ADACC System. In : Neural Information Processing. Springer Berlin Heidelberg, 2013. p. 595-604.

Author:

Ghazal Jaber (ghazal.jaber@gmail.com)

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
protected class	DACC.Pair This helper class is used to sort an array of pairs of integers: val and index.

Field Summary

Fields

Modifier and Type	Field and Description
MultiChoiceOption	combinationOption Combination functions: MAX and WVD (MAX leads to a faster reactivity to the change, WVD is more robust to noise)
protected Classifier[]	ensemble Ensemble of classifiers
protected double[]	ensembleAges Age of classifiers (to compare with maturity age)
protected DACC.Pair[]	ensembleWeights Weights of classifiers
protected int[][]	ensembleWindows Evaluation windows (recent classification errors)
FloatOption	evaluationSizeOption Size of the evaluation window for weights computing
ClassOption	learnerOption Base classifier
FloatOption	maturityOption Maturity age of classifiers
FloatOption	memberCountOption Ensemble size
protected int	nbInstances Number of instances from the stream

Fields inherited from class moa.classifiers.AbstractClassifier

classifierRandom, modelContext, randomSeed, randomSeedOption, trainingWeightSeenByModel

Fields inherited from class moa.options.AbstractOptionHandler

classOptionNamesToPreparedObjects, options

Constructor Summary

Constructors

Constructor and Description
DACC()

Method Summary

Methods

Modifier and Type	Method and Description
void	discardModel(int index) Resets a classifier in the ensemble
protected DACC.Pair[]	getHalf(boolean bestHalf) Returns the best (or worst) half of classifiers in the adaptive ensemble.
protected ArrayList<Integer>	getMAXIndexes() Returns the classifiers that vote for the final prediction when the MAX combination function is selected
void	getModelDescription(StringBuilder out, int indent) Returns a string representation of the model.
protected Measurement[]	getModelMeasurementsImpl() Gets the current measurements of this classifier. The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods.
protected int	getNbActiveClassifiers() Returns the number of classifiers used for prediction which includes the adaptive learners and the snapshots in ADACC
protected int	getNbAdaptiveClassifiers() Returns the number of adaptive classifiers in the ensemble which excludes the static snapshots in ADACC
String	getPurposeString() Gets the purpose of this object
Classifier[]	getSubClassifiers() Gets the classifiers of this ensemble.
double[]	getVotesForInstance(weka.core.Instance inst) Predicts the class memberships for a given instance.
protected ArrayList<Integer>	getWVDIndexes() Returns the classifiers that vote for the final prediction when the WVD combination function is selected
protected void	initVariables() Initializes the method variables
boolean	isRandomizable() Gets whether this classifier needs a random seed.
void	resetLearningImpl() Resets this classifier.
protected void	trainAndClassify(weka.core.Instance inst) Receives a training instance from the stream and updates the adaptive classifiers accordingly
void	trainOnInstanceImpl(weka.core.Instance inst) Trains this classifier incrementally using the given instance. The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods.
protected double	updateEvaluationWindow(int index, int val) Updates the evaluation window of a classifier and returns the updated weight value.

Methods inherited from class moa.classifiers.AbstractClassifier

contextIsCompatible, copy, correctlyClassifies, getAttributeNameString, getAWTRenderer, getClassLabelString, getClassNameString, getDescription, getModelContext, getModelMeasurements, getNominalValueString, modelAttIndexToInstanceAttIndex, modelAttIndexToInstanceAttIndex, prepareForUseImpl, resetLearning, setModelContext, setRandomSeed, trainingHasStarted, trainingWeightSeenByModel, trainOnInstance

Methods inherited from class moa.options.AbstractOptionHandler

discoverOptionsViaReflection, getCLICreationString, getOptions, getPreparedClassOption, getPreparedClassOption, prepareClassOptions, prepareForUse, prepareForUse

Methods inherited from class moa.AbstractMOAObject

copy, measureByteSize, measureByteSize, toString

Methods inherited from class java.lang.Object

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

Methods inherited from interface moa.options.OptionHandler

getCLICreationString, getOptions, prepareForUse, prepareForUse

Methods inherited from interface moa.MOAObject

measureByteSize

Field Detail

learnerOption

public ClassOption learnerOption

Base classifier

memberCountOption

public FloatOption memberCountOption

Ensemble size

maturityOption

public FloatOption maturityOption

Maturity age of classifiers

evaluationSizeOption

public FloatOption evaluationSizeOption

Size of the evaluation window for weights computing

combinationOption

public MultiChoiceOption combinationOption

Combination functions: MAX and WVD (MAX leads to a faster reactivity to the change, WVD is more robust to noise)

ensemble

protected Classifier[] ensemble

Ensemble of classifiers

ensembleWeights

protected DACC.Pair[] ensembleWeights

Weights of classifiers

ensembleAges

protected double[] ensembleAges

Age of classifiers (to compare with maturity age)

ensembleWindows

protected int[][] ensembleWindows

Evaluation windows (recent classification errors)

nbInstances

protected int nbInstances

Number of instances from the stream

Constructor Detail

DACC

public DACC()

Method Detail

getPurposeString

public String getPurposeString()

Description copied from interface: OptionHandler

Gets the purpose of this object

Specified by:

getPurposeString in interface OptionHandler

Overrides:

getPurposeString in class AbstractClassifier

Returns:

the string with the purpose of this object

initVariables

protected void initVariables()

Initializes the method variables

resetLearningImpl

public void resetLearningImpl()

Description copied from class: AbstractClassifier

Resets this classifier. It must be similar to starting a new classifier from scratch.

The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods. Note that this will produce compiler errors if not overridden.

Specified by:

resetLearningImpl in class AbstractClassifier

trainOnInstanceImpl

public void trainOnInstanceImpl(weka.core.Instance inst)

Description copied from class: AbstractClassifier

Trains this classifier incrementally using the given instance.

The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods. Note that this will produce compiler errors if not overridden.

Specified by:

trainOnInstanceImpl in class AbstractClassifier

Parameters:

inst - the instance to be used for training

getVotesForInstance

public double[] getVotesForInstance(weka.core.Instance inst)

Description copied from interface: Classifier

Predicts the class memberships for a given instance. If an instance is unclassified, the returned array elements must be all zero.

Parameters:

inst - the instance to be classified

Returns:

an array containing the estimated membership probabilities of the test instance in each class

trainAndClassify

protected void trainAndClassify(weka.core.Instance inst)

Receives a training instance from the stream and updates the adaptive classifiers accordingly

Parameters:

inst - the instance from the stream

discardModel

public void discardModel(int index)

Resets a classifier in the ensemble

Parameters:

index - the index of the classifier in the ensemble

updateEvaluationWindow

protected double updateEvaluationWindow(int index,
                            int val)

Updates the evaluation window of a classifier and returns the updated weight value.

Parameters:

index - the index of the classifier in the ensemble

val - the last evaluation record of the classifier

Returns:

the updated weight value of the classifier

getHalf

protected DACC.Pair[] getHalf(boolean bestHalf)

Returns the best (or worst) half of classifiers in the adaptive ensemble. The best classifiers are used to compute the stability index in ADACC. The worst classifiers are returned in order to select a classifier for deletion.

Parameters:

bestHalf - boolean value set to true (false) if we want to return the best (worst) half of adaptive classifiers.

horizon -

Returns:

an array containing the weight values of the corresponding classifiers and their indexes in the ensemble.

getMAXIndexes

protected ArrayList<Integer> getMAXIndexes()

Returns the classifiers that vote for the final prediction when the MAX combination function is selected

Returns:

the classifiers with the highest weight value

getWVDIndexes

protected ArrayList<Integer> getWVDIndexes()

Returns the classifiers that vote for the final prediction when the WVD combination function is selected

Returns:

the classifiers whose weights lie in the higher half of the ensemble's weight interval.

getNbActiveClassifiers

protected int getNbActiveClassifiers()

Returns the number of classifiers used for prediction which includes the adaptive learners and the snapshots in ADACC

Returns:

the number of classifiers used for prediction

getNbAdaptiveClassifiers

protected int getNbAdaptiveClassifiers()

Returns the number of adaptive classifiers in the ensemble which excludes the static snapshots in ADACC

Returns:

the number of adaptive classifiers

getModelDescription

public void getModelDescription(StringBuilder out,
                       int indent)

Description copied from class: AbstractClassifier

Returns a string representation of the model.

Specified by:

getModelDescription in class AbstractClassifier

Parameters:

out - the stringbuilder to add the description

indent - the number of characters to indent

getModelMeasurementsImpl

protected Measurement[] getModelMeasurementsImpl()

Description copied from class: AbstractClassifier

Gets the current measurements of this classifier.

The reason for ...Impl methods: ease programmer burden by not requiring them to remember calls to super in overridden methods. Note that this will produce compiler errors if not overridden.

Specified by:

getModelMeasurementsImpl in class AbstractClassifier

Returns:

an array of measurements to be used in evaluation tasks

isRandomizable

public boolean isRandomizable()

Description copied from interface: Classifier

Gets whether this classifier needs a random seed. Examples of methods that needs a random seed are bagging and boosting.

Returns:

true if the classifier needs a random seed.

getSubClassifiers

public Classifier[] getSubClassifiers()

Description copied from interface: Classifier

Gets the classifiers of this ensemble. Returns null if this classifier is a single classifier.

Specified by:

getSubClassifiers in interface Classifier

Overrides:

getSubClassifiers in class AbstractClassifier

Returns:

an array of the classifiers of the ensemble