keel.Algorithms.Neural_Networks.EvRBF_CL

Class RbfnPopulation

java.lang.Object

keel.Algorithms.Neural_Networks.EvRBF_CL.RbfnPopulation

public class RbfnPopulation
extends java.lang.Object

Implements a population of Radial basis Function Neural Networks to be evolved with EvRBFN_CL.

Since:

JDK1.5

Version:

0.1

Author:

Written by Victor Manuel Rivas Santos (University of Jaen) 15/08/2007

Constructor Summary

Constructors

Constructor and Description
RbfnPopulation(int _size) Creates a new instance of RbfnPopulation
RbfnPopulation(int _size, double[][] _X, int _ndatos, int _nEnt, int _nSal, double _neuronsPercentage) Creates and initializes a population using a set of data.

Method Summary

Modifier and Type	Method and Description
void	adder(Rbfn _net) Performs the DELETER mutator operator: modifies C_DELETER% of the Radius of the net
void	applyOperators(double _xOverRate, double _mutatorRate) Applies operator according to their probabilities
void	c_random(Rbfn _net) Performs the C_RANDOM mutator operator: modifies MUTATORS_INTERNAL_PROB % of the centers of the net
void	deleter(Rbfn _net) Performs the DELETER mutator operator: modifies C_DELETER% of the Radius of the net
Rbfn	individual(int _num) Returns the _num-th individuals
void	paint_sort(int _size) Shows the _size first individuals (sorted by fitness)
void	paint() Prints the ppulation on a stdout
void	paint(java.lang.String _fileName) Prints the pipulation on a file.
void	r_random(Rbfn _net) Performs the R_RANDOM mutator operator: modifies 50% of the Radius of the net
void	replaceIndividuals(int _numIndividuals) Removes worse _numIndividuals nets from Population and includes individuals from subpopulation
void	selectIndividuals(int _subPopSize, int _tournamentSize) Creates a subPopulation of individuals using the tournament method.
void	set_min_max_values(double[][] _X, int _ndatos, int _nEnt) Setting the mix and max values for data (needed for operators)
void	setFitness_Cl_subPop(double[][] _X, double[][] _Y, int _nDatos, int _nClases) Sets the fitness of a Sub-Population of RBFNs for classification problems
void	setFitness_Cl(double[][] _X, double[][] _Y, int _nDatos, int _nClases) Sets the fitness of a Population of RBFNs for classification problems
void	sort_population() Sorts population
void	trainLMS_subPop(double[][] _X, double[][] _Y, int _nDatos, int _LMSLoops, double _delta) Trains a Population of RBFNs
void	trainLMS(double[][] _X, double[][] _Y, int _nDatos, int _LMSLoops, double _delta) Trains a Population of RBFNs
void	x_fix(Rbfn _net) Performs the X_FIX crossover operator: replaces numNeurons neurons from _net, taking numNeurons from a randomly chosen net.

Methods inherited from class java.lang.Object

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

Constructor Detail

RbfnPopulation

public RbfnPopulation(int _size)

Creates a new instance of RbfnPopulation

Parameters:

_size - Size of population

RbfnPopulation

public RbfnPopulation(int _size,
                      double[][] _X,
                      int _ndatos,
                      int _nEnt,
                      int _nSal,
                      double _neuronsPercentage)

Creates and initializes a population using a set of data.

Parameters:

_size - Number of nets in the population

_X - Set of input patterns

_ndatos - Number of input patterns

_nEnt - Dimension of input space

_nSal - Dimension of output space

_neuronsPercentage - Percentage of _ndatos used as upper boundary for the number of neurons

Method Detail

set_min_max_values

public void set_min_max_values(double[][] _X,
                               int _ndatos,
                               int _nEnt)

Setting the mix and max values for data (needed for operators)

Parameters:

_X - Data

_ndatos - Number of data (could differ from _X.length)

_nEnt - Input dimension

trainLMS

public void trainLMS(double[][] _X,
                     double[][] _Y,
                     int _nDatos,
                     int _LMSLoops,
                     double _delta)

Trains a Population of RBFNs

Parameters:

_X - Set ot input patterns

_Y - Set of output values

_nDatos - Number of patterns in _X

_LMSLoops - NUmber of iterations for LMS

_delta - Delta parameter for LMS

setFitness_Cl

public void setFitness_Cl(double[][] _X,
                          double[][] _Y,
                          int _nDatos,
                          int _nClases)

Sets the fitness of a Population of RBFNs for classification problems

Parameters:

_X - array containing the input patterns

_Y - array containing the desired output (although only first column is used)

_nDatos - Number of patterns

_nClases - Number of diferent classes

trainLMS_subPop

public void trainLMS_subPop(double[][] _X,
                            double[][] _Y,
                            int _nDatos,
                            int _LMSLoops,
                            double _delta)

Trains a Population of RBFNs

Parameters:

_X - Set ot input patterns

_Y - Set of output values

_nDatos - Number of patterns in _X

_LMSLoops - NUmber of iterations for LMS

_delta - Delta parameter for LMS

setFitness_Cl_subPop

public void setFitness_Cl_subPop(double[][] _X,
                                 double[][] _Y,
                                 int _nDatos,
                                 int _nClases)

Sets the fitness of a Sub-Population of RBFNs for classification problems

Parameters:

_X - array containing the input patterns

_Y - array containing the desired output (although only first column is used)

_nDatos - Number of patterns

_nClases - Number of diferent classes

selectIndividuals

public void selectIndividuals(int _subPopSize,
                              int _tournamentSize)

Creates a subPopulation of individuals using the tournament method.

Parameters:

_subPopSize - Number of individuals to select

_tournamentSize - Number of individuals to perform the tournament

applyOperators

public void applyOperators(double _xOverRate,
                           double _mutatorRate)

Applies operator according to their probabilities

Parameters:

_xOverRate - Probability for XOver operators

_mutatorRate - Probability for Mutation operators

replaceIndividuals

public void replaceIndividuals(int _numIndividuals)

Removes worse _numIndividuals nets from Population and includes individuals from subpopulation

Parameters:

_numIndividuals - Number of indivuals to replace

paint_sort

public void paint_sort(int _size)

Shows the _size first individuals (sorted by fitness)

Parameters:

_size - Number of individuals to show

sort_population

public void sort_population()

Sorts population

x_fix

public void x_fix(Rbfn _net)

Performs the X_FIX crossover operator: replaces numNeurons neurons from _net, taking numNeurons from a randomly chosen net.

Parameters:

_net - The net to be modified

c_random

public void c_random(Rbfn _net)

Performs the C_RANDOM mutator operator: modifies MUTATORS_INTERNAL_PROB % of the centers of the net

Parameters:

_net - The net to be modified

r_random

public void r_random(Rbfn _net)

Performs the R_RANDOM mutator operator: modifies 50% of the Radius of the net

Parameters:

_net - The net to be modified

deleter

public void deleter(Rbfn _net)

Performs the DELETER mutator operator: modifies C_DELETER% of the Radius of the net

Parameters:

_net - The net to be modified

adder

public void adder(Rbfn _net)

Performs the DELETER mutator operator: modifies C_DELETER% of the Radius of the net

Parameters:

_net - The net to be modified

paint

public void paint()

Prints the ppulation on a stdout

paint

public void paint(java.lang.String _fileName)

Prints the pipulation on a file.

Parameters:

_fileName - Name of the file.

individual

public Rbfn individual(int _num)

Returns the _num-th individuals

Parameters:

_num - Position of the individual to return

Returns:

The _num-th individual of the population