mulan.classifier.transformation
Class MultiLabelStacking

java.lang.Object
  mulan.classifier.MultiLabelLearnerBase
      mulan.classifier.transformation.TransformationBasedMultiLabelLearner
          mulan.classifier.transformation.MultiLabelStacking

All Implemented Interfaces:

Serializable, MultiLabelLearner, TechnicalInformationHandler

public class MultiLabelStacking
extends TransformationBasedMultiLabelLearner
implements Serializable

This class is an implementation of the (BR)^2 or Multi-Label stacking method.

For more information, see

Grigorios Tsoumakas, Anastasios Dimou, Eleftherios Spyromitros, Vasileios Mezaris, Ioannis Kompatsiaris, Ioannis Vlahavas: Correlation-Based Pruning of Stacked Binary Relevance Models for Multi-Label Learning. In: Proc. ECML/PKDD 2009 Workshop on Learning from Multi-Label Data (MLD'09), 101-116, 2009.

BibTeX:

 @inproceedings{GrigoriosTsoumakas2009,
    author = {Grigorios Tsoumakas, Anastasios Dimou, Eleftherios Spyromitros, Vasileios Mezaris, Ioannis Kompatsiaris, Ioannis Vlahavas},
    booktitle = {Proc. ECML/PKDD 2009 Workshop on Learning from Multi-Label Data (MLD'09)},
    pages = {101-116},
    title = {Correlation-Based Pruning of Stacked Binary Relevance Models for Multi-Label Learning},
    year = {2009},
    location = {Bled, Slovenia}
 }
 

Version:

2012.02.07

Author:

Eleftherios Spyromitros-Xioufis (espyromi@csd.auth.gr)

See Also:

Serialized Form

Field Summary

protected Instances

train the training instances

Fields inherited from class mulan.classifier.transformation.TransformationBasedMultiLabelLearner

baseClassifier

Fields inherited from class mulan.classifier.MultiLabelLearnerBase

featureIndices, labelIndices, numLabels

Constructor Summary

MultiLabelStacking()
An empty constructor

MultiLabelStacking(Classifier baseClassifier, Classifier metaClassifier)
A constructor with 2 arguments

Method Summary

protected Instances	attachIndexes(Instances original) Attaches an index attribute at the beginning of each instance
void	buildBaseLevel(MultiLabelInstances trainingSet) Builds the base-level classifiers.
void	buildBaseLevelKNN(MultiLabelInstances trainingSet) Used only in case of a kNN base classifier.
protected void	buildInternal(MultiLabelInstances dataSet) Learner specific implementation of building the model from MultiLabelInstances training data set.
void	buildMetaLevel() Builds the ensemble of meta-level classifiers.
protected Instances	detachIndexes(Instances original) Detaches the index attribute from the beginning of each instance
TechnicalInformation	getTechnicalInformation() Returns an instance of a TechnicalInformation object, containing detailed information about the technical background of this class, e.g., paper reference or book this class is based on.
int	getTopkCorrelated() Returns top k correlated labels
String	globalInfo() Returns a string describing the classifier.
void	initializeMetaLevel(MultiLabelInstances dataSet, Classifier metaClassifier, boolean includeAttrs, double metaPercentage, ASEvaluation eval) Initializes all the parameters used in the meta-level.
protected MultiLabelOutput	makePredictionInternal(Instance instance) Learner specific implementation for predicting on specified data based on trained model.
void	saveObject(String filename) Saves a MultiLabelStacking object in a file
void	setEval(ASEvaluation eval) Sets the attribute selection evaluation class
void	setIncludeAttrs(boolean includeAttrs) Sets the value of includeAttrs
void	setMetaAlgorithm(Classifier metaClassifier) Sets the type of the meta classifier and initializes the ensemble
void	setMetaPercentage(double metaPercentage) Sets the value of metaPercentage
void	setNormalize(boolean normalize) Sets the value of normalize
void	setPartialBuild(boolean partialBuild) sets the value for partialBuild

Methods inherited from class mulan.classifier.transformation.TransformationBasedMultiLabelLearner

getBaseClassifier

Methods inherited from class mulan.classifier.MultiLabelLearnerBase

build, debug, getDebug, isModelInitialized, isUpdatable, makeCopy, makePrediction, setDebug

Methods inherited from class java.lang.Object

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

Field Detail

train

protected Instances train

the training instances

Constructor Detail

MultiLabelStacking

public MultiLabelStacking()

An empty constructor

MultiLabelStacking

public MultiLabelStacking(Classifier baseClassifier,
                          Classifier metaClassifier)

A constructor with 2 arguments

Parameters:

baseClassifier - the classifier used in the base-level

metaClassifier - the classifier used in the meta-level

Method Detail

globalInfo

public String globalInfo()

Returns a string describing the classifier.

Overrides:

globalInfo in class TransformationBasedMultiLabelLearner

Returns:

a string description of the classifier

getTechnicalInformation

public TechnicalInformation getTechnicalInformation()

Description copied from class: TransformationBasedMultiLabelLearner

Returns an instance of a TechnicalInformation object, containing detailed information about the technical background of this class, e.g., paper reference or book this class is based on.

Specified by:

getTechnicalInformation in interface TechnicalInformationHandler

Overrides:

getTechnicalInformation in class TransformationBasedMultiLabelLearner

Returns:

the technical information about this class

buildInternal

protected void buildInternal(MultiLabelInstances dataSet)
                      throws Exception

Description copied from class: MultiLabelLearnerBase

Learner specific implementation of building the model from MultiLabelInstances training data set. This method is called from MultiLabelLearnerBase.build(MultiLabelInstances) method, where behavior common across all learners is applied.

Specified by:

buildInternal in class MultiLabelLearnerBase

Parameters:

dataSet - the training data set

Throws:

Exception - if learner model was not created successfully

initializeMetaLevel

public void initializeMetaLevel(MultiLabelInstances dataSet,
                                Classifier metaClassifier,
                                boolean includeAttrs,
                                double metaPercentage,
                                ASEvaluation eval)
                         throws Exception

Initializes all the parameters used in the meta-level. Calculates the correlated labels if meta-level pruning is applied.

Parameters:

dataSet -

metaClassifier -

includeAttrs -

metaPercentage -

eval -

Throws:

Exception

buildBaseLevel

public void buildBaseLevel(MultiLabelInstances trainingSet)
                    throws Exception

Builds the base-level classifiers. Their predictions are gathered in the baseLevelPredictions member

Parameters:

trainingSet -

Throws:

Exception

buildMetaLevel

public void buildMetaLevel()
                    throws Exception

Builds the ensemble of meta-level classifiers.

Throws:

Exception

buildBaseLevelKNN

public void buildBaseLevelKNN(MultiLabelInstances trainingSet)
                       throws Exception

Used only in case of a kNN base classifier.

Parameters:

trainingSet -

Throws:

Exception

makePredictionInternal

protected MultiLabelOutput makePredictionInternal(Instance instance)
                                           throws Exception

Description copied from class: MultiLabelLearnerBase

Learner specific implementation for predicting on specified data based on trained model. This method is called from MultiLabelLearnerBase.makePrediction(weka.core.Instance) which guards for model initialization and apply common handling/behavior.

Specified by:

makePredictionInternal in class MultiLabelLearnerBase

Parameters:

instance - the data instance to predict on

Returns:

the output of the learner for the given instance

Throws:

Exception - if an error occurs while making the prediction.

InvalidDataException - if specified instance data is invalid and can not be processed by the learner

attachIndexes

protected Instances attachIndexes(Instances original)

Attaches an index attribute at the beginning of each instance

Parameters:

original -

Returns:

the modified instances

detachIndexes

protected Instances detachIndexes(Instances original)
                           throws Exception

Detaches the index attribute from the beginning of each instance

Parameters:

original -

Returns:

the modified instances

Throws:

Exception

saveObject

public void saveObject(String filename)

Saves a MultiLabelStacking object in a file

Parameters:

filename -

setNormalize

public void setNormalize(boolean normalize)

Sets the value of normalize

Parameters:

normalize -

setIncludeAttrs

public void setIncludeAttrs(boolean includeAttrs)

Sets the value of includeAttrs

Parameters:

includeAttrs -

setMetaPercentage

public void setMetaPercentage(double metaPercentage)

Sets the value of metaPercentage

Parameters:

metaPercentage -

setEval

public void setEval(ASEvaluation eval)

Sets the attribute selection evaluation class

Parameters:

eval -

setMetaAlgorithm

public void setMetaAlgorithm(Classifier metaClassifier)
                      throws Exception

Sets the type of the meta classifier and initializes the ensemble

Parameters:

metaClassifier -

Throws:

Exception

setPartialBuild

public void setPartialBuild(boolean partialBuild)

sets the value for partialBuild

Parameters:

partialBuild -

getTopkCorrelated

public int getTopkCorrelated()

Returns top k correlated labels

Returns:

top k correlated labels