Serialized Form

Package mulan.classifier

Class mulan.classifier.InvalidDataException extends LearnerException implements Serializable

serialVersionUID: 7578484321526377683L

Class mulan.classifier.LearnerException extends MulanRuntimeException implements Serializable

serialVersionUID: -1893618833783062666L

Class mulan.classifier.ModelInitializationException extends LearnerException implements Serializable

serialVersionUID: 4610550344517098930L

Class mulan.classifier.MultiLabelLearnerBase extends Object implements Serializable

Serialized Fields

isModelInitialized

boolean isModelInitialized

numLabels

int numLabels

The number of labels the learner can handle. The number of labels are determined form the training data when learner is build.

labelIndices

int[] labelIndices

An array containing the indexes of the label attributes within the Instances object of the training data in increasing order. The same order will be followed in the arrays of predictions given by each learner in the MultiLabelOutput object.

featureIndices

int[] featureIndices

An array containing the indexes of the feature attributes within the Instances object of the training data in increasing order.

isDebug

boolean isDebug

Whether debugging is on/off

Package mulan.classifier.lazy

Class mulan.classifier.lazy.BRkNN extends MultiLabelKNN implements Serializable

Serialized Fields

random

Random random

avgPredictedLabels

int avgPredictedLabels

Stores the average number of labels among the knn for each instance Used in BRkNN-b extension

cvMaxK

int cvMaxK

The value of kNN provided by the user. This may differ from numOfNeighbors if cross-validation is being used.

cvkSelection

boolean cvkSelection

Whether to select k by cross validation.

extension

BRkNN.ExtensionType extension

The type of extension to be used

Class mulan.classifier.lazy.IBLR_ML extends MultiLabelKNN implements Serializable

serialVersionUID: 1L

Serialized Fields

classifier

Classifier[] classifier

For each label we create a corresponding binary classifier.

addFeatures

boolean addFeatures

By default, IBLR-ML is used. One can change to IBLR-ML+ through the constructor

Class mulan.classifier.lazy.MLkNN extends MultiLabelKNN implements Serializable

Serialized Fields

smooth

double smooth

Smoothing parameter controlling the strength of uniform prior
(Default value is set to 1 which yields the Laplace smoothing).

PriorProbabilities

double[] PriorProbabilities

A table holding the prior probability for an instance to belong in each class

PriorNProbabilities

double[] PriorNProbabilities

A table holding the prior probability for an instance not to belong in each class

CondProbabilities

double[][] CondProbabilities

A table holding the probability for an instance to belong in each class
given that i:0..k of its neighbors belong to that class

CondNProbabilities

double[][] CondNProbabilities

A table holding the probability for an instance not to belong in each class
given that i:0..k of its neighbors belong to that class

Class mulan.classifier.lazy.MultiLabelKNN extends MultiLabelLearnerBase implements Serializable

Serialized Fields

distanceWeighting

int distanceWeighting

Whether the neighbors should be distance-weighted.

numOfNeighbors

int numOfNeighbors

Number of neighbors used in the k-nearest neighbor algorithm

lnn

LinearNNSearch lnn

Class implementing the brute force search algorithm for nearest neighbor search. Default value is true.

dfunc

DistanceFunction dfunc

Implementing Euclidean distance (or similarity) function.

train

Instances train

The training instances

Package mulan.classifier.meta

Class mulan.classifier.meta.ClusteringBased extends MultiLabelMetaLearner implements Serializable

Serialized Fields

numClusters

int numClusters

The number of clusters

multi

MultiLabelLearner[] multi

The multi-label learners, one for each cluster

clusterer

Clusterer clusterer

The clusterer to use

Class mulan.classifier.meta.ConstrainedKMeans extends RandomizableClusterer implements Serializable

serialVersionUID: -3235809600124455376L

Serialized Fields

bucket

ArrayList<E>[] bucket

bucketSize

int bucketSize

maxIterations

int maxIterations

m_ReplaceMissingFilter

ReplaceMissingValues m_ReplaceMissingFilter

replace missing values in training instances

m_NumClusters

int m_NumClusters

number of clusters to generate

m_ClusterCentroids

Instances m_ClusterCentroids

holds the cluster centroids

m_ClusterStdDevs

Instances m_ClusterStdDevs

Holds the standard deviations of the numeric attributes in each cluster

m_ClusterNominalCounts

int[][][] m_ClusterNominalCounts

For each cluster, holds the frequency counts for the values of each nominal attribute

m_ClusterSizes

int[] m_ClusterSizes

The number of instances in each cluster

m_Min

double[] m_Min

attribute min values

m_Max

double[] m_Max

attribute max values

m_Iterations

int m_Iterations

Keep track of the number of iterations completed before convergence

m_squaredErrors

double[] m_squaredErrors

Holds the squared errors for all clusters

Class mulan.classifier.meta.EnsembleOfSubsetLearners extends MultiLabelMetaLearner implements Serializable

Serialized Fields

ensembleModels

SubsetLearner[] ensembleModels

An array of ensemble models

numModels

int numModels

The number of models in the ensemble

threshold

double threshold

The threshold for ensemble voting

dependenceIdentifier

LabelPairsDependenceIdentifier dependenceIdentifier

Defines the type of dependence identification process.

singleLabelLearner

Classifier singleLabelLearner

Base Classifier that will be used for single label training and predictions

selectDiverseModels

boolean selectDiverseModels

Select most different from the highly weighted partitions

useSubsetcache

boolean useSubsetcache

Disable SubsetLearner caching mechanism

seed

int seed

Seed for replication of random experiments

rnd

Random rnd

Random number generator

Class mulan.classifier.meta.HierarchyBuilder extends Object implements Serializable

Serialized Fields

numPartitions

int numPartitions

labelsXMLDoc

Document labelsXMLDoc

method

HierarchyBuilder.Method method

Class mulan.classifier.meta.HMC extends MultiLabelMetaLearner implements Serializable

Serialized Fields

originalMetaData

LabelsMetaData originalMetaData

root

HMCNode root

labelsAndIndices

Map<K,V> labelsAndIndices

NoNodes

long NoNodes

NoClassifierEvals

long NoClassifierEvals

TotalUsedTrainInsts

long TotalUsedTrainInsts

Class mulan.classifier.meta.HMCNode extends MultiLabelMetaLearner implements Serializable

Serialized Fields

nodeName

String nodeName

children

HashSet<E> children

header

Instances header

Class mulan.classifier.meta.HOMER extends MultiLabelMetaLearner implements Serializable

Serialized Fields

numClusters

int numClusters

hmc

HMC hmc

hb

HierarchyBuilder hb

header

Instances header

method

HierarchyBuilder.Method method

m

MultiLabelInstances m

numMetaLabels

int numMetaLabels

Class mulan.classifier.meta.MultiLabelMetaLearner extends MultiLabelLearnerBase implements Serializable

Serialized Fields

baseLearner

MultiLabelLearner baseLearner

The encapsulated classifier or used for making clones in the case of ensemble classifiers.

Class mulan.classifier.meta.RAkEL extends MultiLabelMetaLearner implements Serializable

Serialized Fields

seed

int seed

Seed for replication of random experiments

rnd

Random rnd

Random number generator

sumVotesIncremental

double[][] sumVotesIncremental

If true then the confidence of the base classifier to the decisions...

lengthVotesIncremental

double[][] lengthVotesIncremental

sumVotes

double[] sumVotes

lengthVotes

double[] lengthVotes

numOfModels

int numOfModels

threshold

double threshold

sizeOfSubset

int sizeOfSubset

classIndicesPerSubset

int[][] classIndicesPerSubset

absoluteIndicesToRemove

int[][] absoluteIndicesToRemove

subsetClassifiers

MultiLabelLearner[] subsetClassifiers

remove

Remove[] remove

combinations

HashSet<E> combinations

Class mulan.classifier.meta.RAkELd extends MultiLabelMetaLearner implements Serializable

Serialized Fields

seed

int seed

Seed for replication of random experiments

rnd

Random rnd

Random number generator

numOfModels

int numOfModels

sizeOfSubset

int sizeOfSubset

classIndicesPerSubset_d

ArrayList<E>[] classIndicesPerSubset_d

absoluteIndicesToRemove

ArrayList<E>[] absoluteIndicesToRemove

listOfLabels

ArrayList<E> listOfLabels

subsetClassifiers

MultiLabelLearner[] subsetClassifiers

remove

Remove[] remove

Class mulan.classifier.meta.SubsetLearner extends MultiLabelMetaLearner implements Serializable

Serialized Fields

multiLabelLearners

ArrayList<E> multiLabelLearners

Arraylist containing the MultiLabelLearners that we will train and use to make the predictions

singleLabelLearners

ArrayList<E> singleLabelLearners

Arraylist containing the FilteredClassifiers that we will train and use to make the predictions

splitOrder

int[][] splitOrder

Array containing the way the labels will be split

absoluteIndicesToRemove

int[][] absoluteIndicesToRemove

Array containing the indices of the labels we are going to remove

remove

Remove[] remove

Array containing the Remove objects used to remove the labels for each split

baseSingleLabelClassifier

Classifier baseSingleLabelClassifier

Base single-label classifier that will be used for training and predictions

useCache

boolean useCache

indication for disabled caching mechanism

clusterer

LabelClustering clusterer

The method used to cluster the labels

Package mulan.classifier.meta.thresholding

Class mulan.classifier.meta.thresholding.ExampleBasedFMeasureOptimizer extends MultiLabelMetaLearner implements Serializable

Class mulan.classifier.meta.thresholding.Meta extends MultiLabelMetaLearner implements Serializable

Serialized Fields

classifier

Classifier classifier

the classifier to learn the number of top labels or the threshold

classifierInstances

Instances classifierInstances

the training instances for the single-label model

metaDatasetChoice

String metaDatasetChoice

the type for constructing the meta dataset

kFoldsCV

int kFoldsCV

the number of folds for cross validation

foldLearner

MultiLabelLearner foldLearner

clean multi-label learner for cross-validation

Class mulan.classifier.meta.thresholding.MetaLabeler extends Meta implements Serializable

Serialized Fields

classChoice

String classChoice

the type of the class

Class mulan.classifier.meta.thresholding.MLPTO extends MultiLabelMetaLearner implements Serializable

Serialized Fields

EBBM

ExampleBasedBipartitionMeasureBase EBBM

Class mulan.classifier.meta.thresholding.OneThreshold extends MultiLabelMetaLearner implements Serializable

Serialized Fields

threshold

double threshold

final threshold value

measure

BipartitionMeasureBase measure

measure for auto-tuning the threshold

folds

int folds

the folds of the cv to evaluate different thresholds

foldLearner

MultiLabelLearner foldLearner

copy of a clean multi-label learner to use at each fold

Class mulan.classifier.meta.thresholding.RCut extends MultiLabelMetaLearner implements Serializable

Serialized Fields

t

int t

the top t number of labels to consider relevant

measure

BipartitionMeasureBase measure

measure for auto-tuning the threshold

folds

int folds

the folds of the cv to evaluate different thresholds

foldLearner

MultiLabelLearner foldLearner

copy of a clean multi-label learner to use at each fold

Class mulan.classifier.meta.thresholding.SCut extends MultiLabelMetaLearner implements Serializable

Serialized Fields

measure

BipartitionMeasureBase measure

measure for auto-tuning the threshold

kFoldsCV

int kFoldsCV

the folds of the cv to evaluate different thresholds

thresholds

double[] thresholds

one threshold for each label to consider relevant

Class mulan.classifier.meta.thresholding.ThresholdPrediction extends Meta implements Serializable

Package mulan.classifier.neural

Class mulan.classifier.neural.BPMLL extends MultiLabelLearnerBase implements Serializable

serialVersionUID: 2153814250172139021L

Serialized Fields

nominalToBinaryFilter

NominalToBinary nominalToBinaryFilter

epochs

int epochs

randomnessSeed

Long randomnessSeed

weightsDecayCost

double weightsDecayCost

learningRate

double learningRate

hiddenLayersTopology

int[] hiddenLayersTopology

normalizeAttributes

boolean normalizeAttributes

normalizer

NormalizationFilter normalizer

model

NeuralNet model

thresholdF

ThresholdFunction thresholdF

Class mulan.classifier.neural.MMPLearner extends MultiLabelLearnerBase implements Serializable

serialVersionUID: 2221778416856852684L

Serialized Fields

perceptrons

List<E> perceptrons

List of tempPerceptrons representing model of the learner. One for each label. They are ordered in same sequence as labels observed from training data.

normalizer

NormalizationFilter normalizer

epochs

int epochs

The number of training epochs to perform with trainig data during the model learning / building

convertNomToBin

boolean convertNomToBin

Indicates whether any nominal attributes from input data set has to be converted to binary

nomToBinFilter

NominalToBinary nomToBinFilter

Filter used for conversion of nominal attributes to binary (if enabled)

lossFunction

RankingLossFunction lossFunction

The measure to be used to judge the performance of ranking when learning the model

mmpUpdateRule

MMPUpdateRuleType mmpUpdateRule

The name of a model update rule used to update the model when learning from training data

isInitialized

boolean isInitialized

The flag indicating if initialization with of learner first learning data samples already took place. This is because the MMPLearner is online and updatable.

randomnessSeed

Long randomnessSeed

Class mulan.classifier.neural.NormalizationFilter extends Object implements Serializable

serialVersionUID: -2575012048861337275L

Serialized Fields

maxValue

double maxValue

minValue

double minValue

attStats

Hashtable<K,V> attStats

Class mulan.classifier.neural.ThresholdFunction extends Object implements Serializable

serialVersionUID: 5347411552628371402L

Serialized Fields

parameters

double[] parameters

Package mulan.classifier.neural.model

Class mulan.classifier.neural.model.ActivationFunction extends Object implements Serializable

Class mulan.classifier.neural.model.ActivationLinear extends ActivationFunction implements Serializable

serialVersionUID: 4255801421493489832L

Class mulan.classifier.neural.model.ActivationTANH extends ActivationFunction implements Serializable

serialVersionUID: -8707244320811304601L

Class mulan.classifier.neural.model.BasicNeuralNet extends Object implements Serializable

serialVersionUID: -8944873770650464701L

Serialized Fields

layers

List<E> layers

currentNetOutput

double[] currentNetOutput

netInputDim

int netInputDim

netOutputDim

int netOutputDim

Class mulan.classifier.neural.model.Neuron extends Object implements Serializable

serialVersionUID: -2826468439369586864L

Serialized Fields

inputWeights

double[] inputWeights

deltaValues

double[] deltaValues

errorValue

double errorValue

function

ActivationFunction function

biasInput

double biasInput

neuronInput

double neuronInput

neuronOutput

double neuronOutput

nextNeurons

List<E> nextNeurons

inputDim

int inputDim

random

Random random

Package mulan.classifier.transformation

Class mulan.classifier.transformation.AdaBoostMH extends IncludeLabelsClassifier implements Serializable

Class mulan.classifier.transformation.BinaryRelevance extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

ensemble

Classifier[] ensemble

The ensemble of binary relevance models. These are Weka Classifier objects.

correspondence

String[] correspondence

The correspondence between ensemble models and labels

brt

BinaryRelevanceTransformation brt

Class mulan.classifier.transformation.CalibratedLabelRanking extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

oneVsOneModels

Classifier[] oneVsOneModels

array holding the one vs one models

numModels

int numModels

number of one vs one models

trainingdata

Instances trainingdata

temporary training data for each one vs one model

metaDataTest

Instances[] metaDataTest

headers of the training sets of the one vs one models

virtualLabelModels

BinaryRelevance virtualLabelModels

binary relevance models for the virtual label

useStandardVoting

boolean useStandardVoting

whether to use standard voting or the fast qweighted algorithm

nodata

boolean[] nodata

whether no data exist for one-vs-one learning

Class mulan.classifier.transformation.ClassifierChain extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

chain

int[] chain

The new chain ordering of the label indices

ensemble

FilteredClassifier[] ensemble

The ensemble of binary relevance models. These are Weka FilteredClassifier objects, where the filter corresponds to removing all label apart from the one that serves as a target for the corresponding model.

Class mulan.classifier.transformation.EnsembleOfClassifierChains extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

numOfModels

int numOfModels

The number of classifier chain models

ensemble

ClassifierChain[] ensemble

An array of ClassifierChain models

rand

Random rand

Random number generator

useConfidences

boolean useConfidences

Whether the output is computed based on the average votes or on the average confidences

useSamplingWithReplacement

boolean useSamplingWithReplacement

Whether to use sampling with replacement to create the data of the models of the ensemble

BagSizePercent

int BagSizePercent

The size of each bag sample, as a percentage of the training size. Used when useSamplingWithReplacement is true

samplingPercentage

double samplingPercentage

The size of each sample, as a percentage of the training size Used when useSamplingWithReplacement is false

Class mulan.classifier.transformation.EnsembleOfPrunedSets extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

threshold

double threshold

Parameter for the threshold of discretization of prediction output

numOfModels

int numOfModels

Parameter for the number of models that constitute the ensemble

percentage

double percentage

Percentage of data

ensemble

PrunedSets[] ensemble

The models in the ensemble

rand

Random rand

Random number generator

Class mulan.classifier.transformation.IncludeLabelsClassifier extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

pt6Trans

IncludeLabelsTransformation pt6Trans

The transformation used by the classifier

transformed

Instances transformed

A dataset with the format needed by the base classifier. It is potentially expensive copying datasets with many attributes, so it is used for building the classifier and then it's mlData are discarded and it is reused during prediction.

Class mulan.classifier.transformation.LabelPowerset extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

confidenceCalculationMethod

int confidenceCalculationMethod

The confidence values for each label are calculated in the following ways 0: Confidence 0 1/0 for all labels, (1 if label true, 0 if label is false) 1: Confidence of x/(1-x) for all labels, where x is the probability of the winning class (x if label true, (1-x) if label is false) 2: Confidence calculated based on the distribution of probabilities obtained from the base classifier, as introduced by the PPT algorithm

makePredictionsBasedOnConfidences

boolean makePredictionsBasedOnConfidences

Whether the method introduced by the PPT algorithm will be used to actually get the 1/0 output bipartition based on the confidences (requires a threshold)

threshold

double threshold

Threshold used for deciding the 1/0 output value of each label based on the corresponding confidences as calculated by the method introduced in the PPT algorithm

transformation

LabelPowersetTransformation transformation

The object that performs the data transformation

Rand

Random Rand

Random number generator for randomly solving tied predictions

Class mulan.classifier.transformation.LabelsetPruning extends LabelPowerset implements Serializable

Serialized Fields

ListInstancePerLabel

HashMap<K,V> ListInstancePerLabel

labelsets and a list with the corresponding instances

p

int p

parameter for the threshold of number of occurences of a labelset

format

Instances format

format of the data

Class mulan.classifier.transformation.MultiClassLearner extends TransformationBasedMultiLabelLearner implements Serializable

Serialized Fields

header

Instances header

transformation

MultiClassTransformation transformation

Class mulan.classifier.transformation.MultiLabelStacking extends TransformationBasedMultiLabelLearner implements Serializable

serialVersionUID: 1L

Serialized Fields

metaClassifier

Classifier metaClassifier

the type of the classifier used in the meta-level

baseLevelData

Instances[] baseLevelData

the BR transformed datasets of the original dataset

metaLevelData

Instances[] metaLevelData

the BR transformed datasets of the meta dataset

baseLevelEnsemble

Classifier[] baseLevelEnsemble

the ensemble of BR classifiers of the original dataset

metaLevelEnsemble

Classifier[] metaLevelEnsemble

the ensemble of BR classifiers of the meta dataset

metaLevelFilteredEnsemble

FilteredClassifier[] metaLevelFilteredEnsemble

the ensemble of pruned BR classifiers of the meta dataset

numFolds

int numFolds

the number of folds used in the first level

train

Instances train

the training instances

baseLevelPredictions

double[][] baseLevelPredictions

a table holding the predictions of the first level classifiers for each class-label of every instance

normalize

boolean normalize

whether to normalize baseLevelPredictions or not.

maxProb

double[] maxProb

a table holding the maximum probability of each label according to the predictions of the base level classifiers

minProb

double[] minProb

a table holding the minimum probability of each label according to the predictions of the base level classifiers

includeAttrs

boolean includeAttrs

whether to include the original attributes in the meta-level

metaPercentage

double metaPercentage

defines the percentage of labels used in the meta-level

topkCorrelated

int topkCorrelated

The number of labels that will be used for training the meta-level classifiers. The value is derived by metaPercentage and used only internally

selectedAttributes

int[][] selectedAttributes

A table holding the attributes of the most correlated labels for each label.

eval

ASEvaluation eval

The attribute selection evaluator used for pruning the meta-level attributes.

lnn

LinearNNSearch lnn

Class implementing the brute force search algorithm for nearest neighbor search. Used only in case of a kNN baseClassifier

partialBuild

boolean partialBuild

Whether base and meta level are going to be built separately. If true then the buildInternal method doesn't build anything.

Class mulan.classifier.transformation.PPT extends LabelsetPruning implements Serializable

Serialized Fields

strategy

PPT.Strategy strategy

strategy for processing infrequent labelsets

Class mulan.classifier.transformation.PrunedSets extends LabelsetPruning implements Serializable

Serialized Fields

strategy

PrunedSets.Strategy strategy

strategy for processing infrequent labelsets

b

int b

parameter of strategy for processing infrequent labelsets

Class mulan.classifier.transformation.TransformationBasedMultiLabelLearner extends MultiLabelLearnerBase implements Serializable

Serialized Fields

baseClassifier

Classifier baseClassifier

The underlying single-label classifier.

Package mulan.core

Class mulan.core.ArgumentNullException extends MulanRuntimeException implements Serializable

serialVersionUID: -555866519789329786L

Serialized Fields

paramName

String paramName

The name of the parameter which caused the exception

Class mulan.core.MulanException extends Exception implements Serializable

serialVersionUID: 2271544759439172440L

Class mulan.core.MulanRuntimeException extends RuntimeException implements Serializable

serialVersionUID: -1043393739740022098L

Class mulan.core.WekaException extends MulanRuntimeException implements Serializable

serialVersionUID: -8041689691825060987L

Package mulan.data

Class mulan.data.ConditionalDependenceIdentifier extends Object implements Serializable

Serialized Fields

criticalValue

double criticalValue

A default t-critical value, corresponds to significance level 0.01. Label pairs with dependence value below the critical are considered as independent.

baseLearner

Classifier baseLearner

A single-label classifier used to perform dependence test between labels.

numFolds

int numFolds

Number of folds used for cross validation.

seed

int seed

Seed for replication of random experiments

Class mulan.data.DataLoadException extends MulanRuntimeException implements Serializable

serialVersionUID: 1102055196162204723L

Class mulan.data.GreedyLabelClustering extends Object implements Serializable

Serialized Fields

singleLabelLearner

Classifier singleLabelLearner

Classifier that will be used for single label training and predictions

multiLabelLearner

MultiLabelLearner multiLabelLearner

Classifier that will be used for multi-label training and predictions

depLabelsIdentifier

LabelPairsDependenceIdentifier depLabelsIdentifier

Defines the type of dependence identification process.

criticalValue

double criticalValue

Critical value below which the label pairs are considered independent. If set to 0 - the critical value returned by LabelPairsDependenceIdentifier.getCriticalValue() is used.

numFolds

int numFolds

Number of folds for evaluation of SubsetLearner models

allowedNonImprovementSteps

int allowedNonImprovementSteps

Number of allowed concurrent steps with reduced accuracy

measure

Measure measure

Measure by which models are compared

useSubsetLearnerCache

boolean useSubsetLearnerCache

Enable SubsetLearner caching mechanism

internalSubsetLearnerDebug

boolean internalSubsetLearnerDebug

Enable debug output of the internal SubsetLearner

Class mulan.data.InvalidDataFormatException extends MulanException implements Serializable

serialVersionUID: -8323657086903118700L

Class mulan.data.LabelNodeImpl extends Object implements Serializable

serialVersionUID: -7974176487751728557L

Serialized Fields

name

String name

childrenNodes

Set<E> childrenNodes

parentNode

LabelNode parentNode

Class mulan.data.LabelsBuilderException extends MulanException implements Serializable

serialVersionUID: 2161709838882541792L

Class mulan.data.LabelSet extends Object implements Serializable

serialVersionUID: 7658871740607834759L

Serialized Fields

labelSet

int[] labelSet

The set is represented internally as an array of integers. Weka uses doubles but we choose integers for fast comparisons. Observe that the set is immutable, once assigned by the constructor, no changes are possible.

size

int size

A cached count of the set size. Observe that the set size is not the same as the size of the double array.

Class mulan.data.LabelsMetaDataImpl extends Object implements Serializable

serialVersionUID: 5098050799557336378L

Serialization Methods

readExternal

public void readExternal(ObjectInput in)
                  throws IOException,
                         ClassNotFoundException

Throws:

IOException

ClassNotFoundException

writeExternal

public void writeExternal(ObjectOutput out)
                   throws IOException

Throws:

IOException

Class mulan.data.LabelsPair extends Object implements Serializable

Serialized Fields

pair

int[] pair

a pair of labels

score

Double score

dependence score of the labels pair

Class mulan.data.MultiLabelInstances extends Object implements Serializable

Serialized Fields

dataSet

Instances dataSet

labelsMetaData

LabelsMetaData labelsMetaData

Class mulan.data.Statistics extends Object implements Serializable

serialVersionUID: 1206845794397561633L

Serialized Fields

numInstances

int numInstances

the number of instances

numPredictors

int numPredictors

the number of predictive attributes

numNominal

int numNominal

the number of nominal predictive attributes

numNumeric

int numNumeric

the number of numeric attributes

numLabels

int numLabels

the number of labels

labelDensity

double labelDensity

the label density

labelCardinality

double labelCardinality

the label cardinality

examplesPerLabel

double[] examplesPerLabel

percentage of instances per label

cardinalityDistribution

double[] cardinalityDistribution

number of examples per cardinality,

note that this array has size equal to the number of elements plus one,
because the first element is the number of examples for cardinality=0

labelsets

HashMap<K,V> labelsets

labelsets and their frequency

phi

double[][] phi

the array holding the phi correlations

Class mulan.data.UnconditionalChiSquareIdentifier extends Object implements Serializable

Serialized Fields

criticalValue

double criticalValue

A default chi square critical value, corresponds to significance level 0.01. Label pairs with dependence value below the critical are considered as independent.

Package mulan.dimensionalityReduction

Class mulan.dimensionalityReduction.BinaryRelevanceAttributeEvaluator extends ASEvaluation implements Serializable

Serialized Fields

scores

double[] scores

final scores for all attributes

numLabels

int numLabels

The number of labels

CombApprMode

String CombApprMode

combination approach mode

NormMode

String NormMode

normalization mode

ScoreMode

String ScoreMode

attribute scoring based either on evaluation scores or ranking

Class mulan.dimensionalityReduction.LabelPowersetAttributeEvaluator extends ASEvaluation implements Serializable

Serialized Fields

baseAttributeEvaluator

ASEvaluation baseAttributeEvaluator

The single-label attribute evaluator to use underneath

Class mulan.dimensionalityReduction.MultiClassAttributeEvaluator extends ASEvaluation implements Serializable

Serialized Fields

baseAttributeEvaluator

ASEvaluation baseAttributeEvaluator

The single-label attribute evaluator to use underneath

Package mulan.evaluation.loss

Class mulan.evaluation.loss.BipartitionLossFunctionBase extends Object implements Serializable

Class mulan.evaluation.loss.ErrorSetSize extends RankingLossFunctionBase implements Serializable

Class mulan.evaluation.loss.HammingLoss extends BipartitionLossFunctionBase implements Serializable

Class mulan.evaluation.loss.HierarchicalLoss extends BipartitionLossFunctionBase implements Serializable

Serialized Fields

metaData

LabelsMetaData metaData

labelPosition

Map<K,V> labelPosition

loss

double loss

Class mulan.evaluation.loss.IsError extends RankingLossFunctionBase implements Serializable

Class mulan.evaluation.loss.OneError extends RankingLossFunctionBase implements Serializable

Class mulan.evaluation.loss.OneMinusAveragePrecision extends RankingLossFunctionBase implements Serializable

Class mulan.evaluation.loss.RankingLoss extends ErrorSetSize implements Serializable

Class mulan.evaluation.loss.RankingLossFunctionBase extends Object implements Serializable

Package mulan.evaluation.measure

Class mulan.evaluation.measure.AveragePrecision extends RankingMeasureBase implements Serializable

Class mulan.evaluation.measure.BipartitionMeasureBase extends MeasureBase implements Serializable

Class mulan.evaluation.measure.ConfidenceMeasureBase extends MeasureBase implements Serializable

Class mulan.evaluation.measure.Coverage extends RankingMeasureBase implements Serializable

Class mulan.evaluation.measure.ErrorSetSize extends LossBasedRankingMeasureBase implements Serializable

Class mulan.evaluation.measure.ExampleBasedAccuracy extends ExampleBasedBipartitionMeasureBase implements Serializable

Serialized Fields

forgivenessRate

double forgivenessRate

Class mulan.evaluation.measure.ExampleBasedBipartitionMeasureBase extends BipartitionMeasureBase implements Serializable

Serialized Fields

sum

double sum

The current sum of the measure

count

int count

The number of validation examples processed

Class mulan.evaluation.measure.ExampleBasedFMeasure extends ExampleBasedBipartitionMeasureBase implements Serializable

Serialized Fields

beta

double beta

Class mulan.evaluation.measure.ExampleBasedPrecision extends ExampleBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.ExampleBasedRecall extends ExampleBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.ExampleBasedSpecificity extends ExampleBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.GeometricMeanAverageInterpolatedPrecision extends MeanAverageInterpolatedPrecision implements Serializable

Class mulan.evaluation.measure.GeometricMeanAveragePrecision extends MeanAveragePrecision implements Serializable

Class mulan.evaluation.measure.HammingLoss extends LossBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.HierarchicalLoss extends LossBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.IsError extends LossBasedRankingMeasureBase implements Serializable

Class mulan.evaluation.measure.LabelBasedAUC extends ConfidenceMeasureBase implements Serializable

Serialized Fields

numOfLabels

int numOfLabels

The number of labels

m_Predictions

FastVector<E>[] m_Predictions

The predictions for each label

all_Predictions

FastVector<E> all_Predictions

The predictions for all labels

Class mulan.evaluation.measure.LabelBasedAveragePrecision extends ConfidenceMeasureBase implements Serializable

Serialized Fields

numOfLabels

int numOfLabels

the number of labels

confact

List<E>[] confact

collection that stores all predictions and ground truths

Class mulan.evaluation.measure.LabelBasedAveragePrecision.ConfidenceActual extends Object implements Serializable

Serialized Fields

actual

boolean actual

confidence

double confidence

Class mulan.evaluation.measure.LabelBasedBipartitionMeasureBase extends BipartitionMeasureBase implements Serializable

Serialized Fields

numOfLabels

int numOfLabels

the number of labels

falseNegatives

double[] falseNegatives

the number of false negative for each label

truePositives

double[] truePositives

the number of true positives for each label

falsePositives

double[] falsePositives

the number of false positives for each label

trueNegatives

double[] trueNegatives

the number of true negatives for each label

Class mulan.evaluation.measure.LabelBasedFMeasure extends LabelBasedBipartitionMeasureBase implements Serializable

Serialized Fields

beta

double beta

the parameter for combining precision and recall

Class mulan.evaluation.measure.LabelBasedPrecision extends LabelBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.LabelBasedRecall extends LabelBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.LabelBasedSpecificity extends LabelBasedBipartitionMeasureBase implements Serializable

Class mulan.evaluation.measure.LossBasedBipartitionMeasureBase extends ExampleBasedBipartitionMeasureBase implements Serializable

Serialized Fields

loss

BipartitionLossFunction loss

Class mulan.evaluation.measure.LossBasedRankingMeasureBase extends RankingMeasureBase implements Serializable

Serialized Fields

loss

RankingLossFunction loss

Class mulan.evaluation.measure.MacroAUC extends LabelBasedAUC implements Serializable

Class mulan.evaluation.measure.MacroFMeasure extends LabelBasedFMeasure implements Serializable

Class mulan.evaluation.measure.MacroPrecision extends LabelBasedPrecision implements Serializable

Class mulan.evaluation.measure.MacroRecall extends LabelBasedRecall implements Serializable

Class mulan.evaluation.measure.MacroSpecificity extends LabelBasedSpecificity implements Serializable

Class mulan.evaluation.measure.MeanAverageInterpolatedPrecision extends LabelBasedAveragePrecision implements Serializable

Serialized Fields

numRecallLevels

int numRecallLevels

The number of recall levels for which MeanAverageInterpolatedPrecision is calculated

Class mulan.evaluation.measure.MeanAveragePrecision extends LabelBasedAveragePrecision implements Serializable

Class mulan.evaluation.measure.MeasureBase extends Object implements Serializable

Class mulan.evaluation.measure.MicroAUC extends LabelBasedAUC implements Serializable

Class mulan.evaluation.measure.MicroFMeasure extends LabelBasedFMeasure implements Serializable

Class mulan.evaluation.measure.MicroPrecision extends LabelBasedPrecision implements Serializable

Class mulan.evaluation.measure.MicroRecall extends LabelBasedRecall implements Serializable

Class mulan.evaluation.measure.MicroSpecificity extends LabelBasedSpecificity implements Serializable

Class mulan.evaluation.measure.OneError extends LossBasedRankingMeasureBase implements Serializable

Class mulan.evaluation.measure.RankingLoss extends LossBasedRankingMeasureBase implements Serializable

Class mulan.evaluation.measure.RankingMeasureBase extends MeasureBase implements Serializable

Serialized Fields

sum

double sum

The current sum of the measure

count

int count

The number of validation examples processed

Class mulan.evaluation.measure.SubsetAccuracy extends ExampleBasedBipartitionMeasureBase implements Serializable

Package mulan.transformations

Class mulan.transformations.BinaryRelevanceTransformation extends Object implements Serializable

Serialized Fields

data

MultiLabelInstances data

shell

Instances shell

remove

Remove remove

add

Add add

Class mulan.transformations.IncludeLabelsTransformation extends Object implements Serializable

Serialized Fields

labelIndices

int[] labelIndices

Class mulan.transformations.LabelPowersetTransformation extends Object implements Serializable

Serialized Fields

transformedFormat

Instances transformedFormat

Package mulan.transformations.multiclass

Class mulan.transformations.multiclass.Copy extends MultiClassTransformationBase implements Serializable

Class mulan.transformations.multiclass.CopyWeight extends Copy implements Serializable

Class mulan.transformations.multiclass.Ignore extends MultiClassTransformationBase implements Serializable

Class mulan.transformations.multiclass.MultiClassTransformationBase extends Object implements Serializable

Serialized Fields

numOfLabels

int numOfLabels

the number of labels

labelIndices

int[] labelIndices

the array with the label indices

Class mulan.transformations.multiclass.SelectBasedOnFrequency extends MultiClassTransformationBase implements Serializable

Serialized Fields

type

SelectionType type

type of frequency

labelOccurance

int[] labelOccurance

occurences of each label

Class mulan.transformations.multiclass.SelectRandom extends MultiClassTransformationBase implements Serializable