keel.Algorithms.Instance_Generation.HYB

Class HYBGenerator

java.lang.Object

keel.Algorithms.Instance_Generation.Basic.PrototypeGenerator

keel.Algorithms.Instance_Generation.HYB.HYBGenerator

public class HYBGenerator
extends PrototypeGenerator

Hybrid algorithm

Author:

diegoj

Field Summary

Fields

Modifier and Type	Field and Description
protected double	alpha_0 Alpha0 parameter of the LVQ3 methods.
static java.lang.String	CNN CNN title text
protected double	deltaE Increment of epsilon in each step.
protected double	deltaW Increment of window width in each step.
protected int	desiredFinalSize Desired final size of the generated prototype set.
protected int	iterations Iterations of the final optimal LVQ3.
protected double	partitionPercentTraining % of training data set used as training in the search process.
protected double	perOneGenerated Per one prototypes generated in the LVQ3.
static java.lang.String	PNN PNNGenerator title text
protected int	searchingIterations Iterations of each LVQ3-search for optimal epsilon and window width.
static java.lang.String	SVM SVM title text
protected java.lang.String	typeOfInitialReduction Type of initial reduction.
static java.lang.String[]	typesOfReduction Titles of types of initial reduction
static java.lang.String	VQ VQGenerator title text

Fields inherited from class keel.Algorithms.Instance_Generation.Basic.PrototypeGenerator

algorithmName, generatedDataSet, SEED, seedDefaultValueList, trainingDataSet

Constructor Summary

Constructors

Constructor and Description
HYBGenerator(PrototypeSet tDataSet, int iterSearch, int iterEnd, double percentageGeneratedByOptimalLVQ3, double alpha, double initW, double endW, double deltaW, double initE, double endE, double deltaE, double percentPartition, java.lang.String type, java.lang.String paramsOfInitialReducction) Construct a HYBGenerator algorithm.
HYBGenerator(PrototypeSet tDataSet, int iterSearch, int iterEnd, double percentageGeneratedByOptimalLVQ3, double alpha, double initW, double endW, double deltaW, double initE, double endE, double deltaE, double percentPartition, java.lang.String type, java.lang.String[] paramsOfInitialReducction) Construct a HYBGenerator algorithm.
HYBGenerator(PrototypeSet _trainingDataSet, Parameters parameters) Constructor of HYBGenerator algorithm.

Method Summary

Modifier and Type	Method and Description
protected PrototypeSet	initialReduction() Makes the initial reduction of the prototype set
static void	main(java.lang.String[] args) General main for all the prototoype generators Arguments: 0: Filename with the training data set to be condensed. 1: Filename wich contains the test data set. 3: Seed of the random generator. 4: Iterations performed in the search of optimal LVQ3-parameters. 5: Iterations performed in optimal LVQ3 process. 6: % of prototypes generated in optimal LVQ3 process. 7: Alpha0 LVQ3-parameter. 8: Initial value of window width. 9: Final value of window width. 10: Increment in each step of the value of window width. 11: Initial value of epsilon. 12: Final value of epsilon. 13: Increment in each step of the value of epsilon. 14: % de initial partition in the training set. 15: Type of initial reduction performed. 16 and so on: Parameters of the initial reduction process
PrototypeSet	reduceSet() Reduce the set by the Hybrid method.

Methods inherited from class keel.Algorithms.Instance_Generation.Basic.PrototypeGenerator

absoluteAccuracy, absoluteAccuracyAndError, absoluteAccuracyKNN, accuracy, accuracy2, desordenar_vector_sin, desordenar_vector, execute, generateReducedDataSet, getResultingAccuracy, getResultingAccuracy, getResults, getResultsOfAccuracy, getResultsOfAccuracy, getSeed, getSetSizeFromPercentage, getSetSizeFromPercentage, getTime, inic_vector_sin, inic_vector, saveResultsOfAccuracyIn, saveResultsOfAccuracyIn, selecRandomSet, setSeed, showResultsOfAccuracy, showResultsOfAccuracy

Methods inherited from class java.lang.Object

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

Field Detail

PNN

public static final java.lang.String PNN

PNNGenerator title text

See Also:

Constant Field Values

SVM

public static final java.lang.String SVM

SVM title text

See Also:

Constant Field Values

CNN

public static final java.lang.String CNN

CNN title text

See Also:

Constant Field Values

VQ

public static final java.lang.String VQ

VQGenerator title text

See Also:

Constant Field Values

typesOfReduction

public static final java.lang.String[] typesOfReduction

Titles of types of initial reduction

deltaE

protected double deltaE

Increment of epsilon in each step.

deltaW

protected double deltaW

Increment of window width in each step.

partitionPercentTraining

protected double partitionPercentTraining

% of training data set used as training in the search process.

desiredFinalSize

protected int desiredFinalSize

Desired final size of the generated prototype set.

searchingIterations

protected int searchingIterations

Iterations of each LVQ3-search for optimal epsilon and window width.

iterations

protected int iterations

Iterations of the final optimal LVQ3.

perOneGenerated

protected double perOneGenerated

Per one prototypes generated in the LVQ3.

alpha_0

protected double alpha_0

Alpha0 parameter of the LVQ3 methods.

typeOfInitialReduction

protected java.lang.String typeOfInitialReduction

Type of initial reduction. Posible values: PNNGenerator, CNN, SVM, VQGenerator.

Constructor Detail

HYBGenerator

public HYBGenerator(PrototypeSet _trainingDataSet,
                    Parameters parameters)

Constructor of HYBGenerator algorithm.

Parameters:

_trainingDataSet - TrainingDataSet.

parameters - Parameters of the algorithm.

HYBGenerator

public HYBGenerator(PrototypeSet tDataSet,
                    int iterSearch,
                    int iterEnd,
                    double percentageGeneratedByOptimalLVQ3,
                    double alpha,
                    double initW,
                    double endW,
                    double deltaW,
                    double initE,
                    double endE,
                    double deltaE,
                    double percentPartition,
                    java.lang.String type,
                    java.lang.String[] paramsOfInitialReducction)

Construct a HYBGenerator algorithm.

Parameters:

tDataSet - Training data set.

iterSearch - Iterations performed in the internals LVQ3.

iterEnd - Iterations performed in the optimal (and final) LVQ3.

percentageGeneratedByOptimalLVQ3 - % of size of initial set to be generated.

alpha - Alpha 0 LVQ3 parameter.

initW - Initial value of window width.

endW - End value of window width.

deltaW - Increment of the window width.

initE - Initial value of epsilon.

endE - End value of epsilon.

deltaE - Increment of the epsilon.

percentPartition - % of training data set used as training in the search process.

type - Type of initial reduction: CNN, VQGenerator, SVM, PNNGenerator.

paramsOfInitialReducction - Parameters of the initial reduction.

HYBGenerator

public HYBGenerator(PrototypeSet tDataSet,
                    int iterSearch,
                    int iterEnd,
                    double percentageGeneratedByOptimalLVQ3,
                    double alpha,
                    double initW,
                    double endW,
                    double deltaW,
                    double initE,
                    double endE,
                    double deltaE,
                    double percentPartition,
                    java.lang.String type,
                    java.lang.String paramsOfInitialReducction)

Construct a HYBGenerator algorithm.

Parameters:

tDataSet - Training data set.

iterSearch - Iterations performed in the internals LVQ3.

iterEnd - Iterations performed in the optimal (and final) LVQ3.

percentageGeneratedByOptimalLVQ3 - % of size of initial set to be generated.

alpha - Alpha 0 LVQ3 parameter.

initW - Initial value of window width.

endW - End value of window width.

deltaW - Increment of the window width.

initE - Initial value of epsilon.

endE - End value of epsilon.

deltaE - Increment of the epsilon.

percentPartition - % of training data set used as training in the search process.

type - Type of initial reduction: CNN, VQ, SVM, PNN.

paramsOfInitialReducction - Parameters of the initial reduction. In form of plain string.

Method Detail

initialReduction

protected PrototypeSet initialReduction()

Makes the initial reduction of the prototype set

Returns:

Reduced prototype set.

reduceSet

public PrototypeSet reduceSet()

Reduce the set by the Hybrid method.

Overrides:

reduceSet in class PrototypeGenerator

Returns:

Prototype set generated by Hybrid method.

main

public static void main(java.lang.String[] args)

General main for all the prototoype generators Arguments: 0: Filename with the training data set to be condensed. 1: Filename wich contains the test data set. 3: Seed of the random generator. 4: Iterations performed in the search of optimal LVQ3-parameters. 5: Iterations performed in optimal LVQ3 process. 6: % of prototypes generated in optimal LVQ3 process. 7: Alpha0 LVQ3-parameter. 8: Initial value of window width. 9: Final value of window width. 10: Increment in each step of the value of window width. 11: Initial value of epsilon. 12: Final value of epsilon. 13: Increment in each step of the value of epsilon. 14: % de initial partition in the training set. 15: Type of initial reduction performed. 16 and so on: Parameters of the initial reduction process

Parameters:

args - Arguments of the main function.