Package unifeat.featureSelection.embedded.SVMBasedMethods

Class SVMBasedMethods

java.lang.Object

unifeat.featureSelection.FeatureSelection

unifeat.featureSelection.embedded.EmbeddedApproach

unifeat.featureSelection.embedded.SVMBasedMethods.SVMBasedMethods

Direct Known Subclasses:

MSVM_RFE, OVA_SVM_RFE, OVO_SVM_RFE, SVM_RFE

public abstract class SVMBasedMethods
extends EmbeddedApproach

The abstract class contains the main methods and fields that are used in all SVM-based feature selection methods. This class inherits from EmbeddedApproach class.

Author:

Sina Tabakhi

See Also:

• EmbeddedApproach
• FeatureSelection

Field Summary

Fields

Modifier and Type	Field	Description
protected double[]	classLabelInTrainSet
protected SVMKernelType	kernelType
protected double	parameterC
protected final String	TEMP_PATH

Fields inherited from class unifeat.featureSelection.embedded.EmbeddedApproach

classLabel, nameFeatures, PROJECT_PATH

Fields inherited from class unifeat.featureSelection.FeatureSelection

numClass, numFeatures, numSelectedFeature, selectedFeatureSubset, trainSet

Constructor Summary

Constructors

Constructor	Description
SVMBasedMethods(Object... arguments)	Initializes the parameters
SVMBasedMethods(String path, SVMKernelType kernelType, double c)	Initializes the parameters

Method Summary

Modifier and Type	Method	Description
protected double[][][]	buildSVM_OneAgainstOne(int[] selectedFeature)	Generates binary classifiers (SVM) using input data and based on selected feature subset, and finally returns the weights of features.
protected double[][]	buildSVM_OneAgainstRest(int[] selectedFeature)	Generates binary classifiers (SVM) using input data and based on selected feature subset, and finally returns the weights of features.
void	createClassLabel()	Creates an array of class labels available in the train set

Methods inherited from class unifeat.featureSelection.embedded.EmbeddedApproach

loadDataSet, loadDataSet, newMethod, originalFeatureSet

Methods inherited from class unifeat.featureSelection.FeatureSelection

evaluateFeatures, getSelectedFeatureSubset, setNumSelectedFeature, validate

Methods inherited from class java.lang.Object

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

Field Details

TEMP_PATH

protected final String TEMP_PATH

kernelType

protected SVMKernelType kernelType

parameterC

protected double parameterC

classLabelInTrainSet

protected double[] classLabelInTrainSet

Constructor Details

SVMBasedMethods

public SVMBasedMethods(Object... arguments)

Initializes the parameters

Parameters:

arguments - array of parameters contains (path, kernelType, Parameter c) in which path is the path of the project, kernelType is the type of kernel to use, and Parameter c is the complexity parameter C

SVMBasedMethods

public SVMBasedMethods(String path,
 SVMKernelType kernelType,
 double c)

Initializes the parameters

Parameters:

path - the path of the project

kernelType - the type of kernel to use

c - the complexity parameter C

Method Details

createClassLabel

public void createClassLabel()

Creates an array of class labels available in the train set

buildSVM_OneAgainstOne

protected double[][][] buildSVM_OneAgainstOne(int[] selectedFeature)

Generates binary classifiers (SVM) using input data and based on selected feature subset, and finally returns the weights of features. One-Versus-One strategy is used to construct classifiers in multiclass classification.

Parameters:

selectedFeature - an array of indices of the selected feature subset

Returns:

an array of the weights of features

buildSVM_OneAgainstRest

protected double[][] buildSVM_OneAgainstRest(int[] selectedFeature)

Generates binary classifiers (SVM) using input data and based on selected feature subset, and finally returns the weights of features. One-Versus-All strategy is used to construct classifiers in multiclass classification.

Parameters:

selectedFeature - an array of indices of the selected feature subset

Returns:

an array of the weights of features