adams.optimise.genetic
Class AbstractGeneticAlgorithm

java.lang.Object
  adams.core.ConsoleObject
      adams.core.option.AbstractOptionHandler
          adams.optimise.AbstractOptimiser
              adams.optimise.genetic.AbstractGeneticAlgorithm

All Implemented Interfaces:

CleanUpHandler, Debuggable, Destroyable, OptionHandler, SizeOfHandler, Serializable, Comparable

Direct Known Subclasses:

PackDataGeneticAlgorithm

public abstract class AbstractGeneticAlgorithm
extends AbstractOptimiser

Base class for genetic algorithms.

Version:

$Revision: 4584 $

Author:

Dale (dale at cs dot waikato dot ac dot nz)

See Also:

Serialized Form

Field Summary

protected Range	m_BestRange stores the best range of attribtues.
protected boolean	m_FavorZeroes whether to favor 0s instead of 1s.
protected double[]	m_Fitness the fitness of the genes.
protected BitSet[]	m_Genes the genes.
protected int	m_Iterations number of iterations.
protected int	m_MaxTrainTime the maximum number of seconds to train.
protected int	m_NumChrom number of chromosomes.
protected int	m_NumGenes number of genes per chromosome.
protected Random	m_Random the random number generator.
protected boolean	m_Running whether the algorithm is still running.
protected int	m_Seed the seed value.
protected long	m_TrainStart the time when training commenced.
static String	PROPS_FILTER the key for a filter setup in the setup properties.
static String	PROPS_MASK the key for the mask in the setup properties.
static String	PROPS_RELATION the key for the relation name in the generated properties file.

Fields inherited from class adams.optimise.AbstractOptimiser

m_bestf, m_bestv

Fields inherited from class adams.core.option.AbstractOptionHandler

m_DebugLevel, m_OptionManager

Constructor Summary

AbstractGeneticAlgorithm()

Method Summary

String	bestRangeTipText() Returns the tip text for this property.
abstract void	calcFitness() Override the following function in sub-classes.
protected void	copyGene(int to, int from) Copies the values of one gene to another.
void	defineOptions() Adds options to the internal list of options.
void	doCrossovers() Performs cross-over.
void	doMutations() Performs mutations.
void	doMutations2() Performs mutations.
String	favorZeroesTipText() Returns the tip text for this property.
static AbstractGeneticAlgorithm	forCommandLine(String cmdline) Instantiates the genetic algorithm from the given commandline (i.e., classname and optional options).
static AbstractGeneticAlgorithm	forName(String classname, String[] options) Instantiates the genetic algorithm with the given options.
String	getBestRange() Returns the best range of attributes.
boolean	getFavorZeroes() Returns whether 0s are favored over 1s.
double[]	getFitness() Returns the current fitness values.
boolean	getGene(int chromosome, int gene) Returns the value of the specified gene.
static String[]	getGeneticAlgorithms() Returns a list with classnames of genetic algorithms.
abstract Vector<int[]>	getInitialSetups()
int	getIterations() Gets the iterations use.
int	getMaxTrainTime() Returns the maximum number of seconds to perform training.
int	getNumChrom() Returns the number of chromosomes to use.
int	getNumGenes() Returns the number of genes to use.
int	getSeed() Returns the current seed value.
boolean	hasMoreZeroes(BitSet a, BitSet b) return if a has more zeroes than b.
protected void	init(int ch, int genes) Initializes the algorithm.
protected void	initialize() Initializes the members.
boolean	isRunning() Returns whether the algorithm is still running.
String	iterationsTipText() Returns the tip text for this property.
String	maxTrainTimeTipText() Returns the tip text for this property.
String	numChromTipText() Returns the tip text for this property.
protected void	postRun() Further clean-ups in derived classes.
protected void	preRun() Further initializations in derived classes.
void	reset() Resets the genetic algorihtm.
boolean	run() Runs the genetic algorithm.
static void	runGeneticAlgorithm(Class env, Class genetic, String[] options) Runs the genetic algorithm with the given options.
String	seedTipText() Returns the tip text for this property.
protected void	setBestRange(Range value) Sets the best range of attributes.
void	setBestRange(String value) Sets the best range of attributes.
void	setFavorZeroes(boolean value) Sets whether 0s are favored over 1s.
void	setGene(int chromosome, int gene, boolean value) Sets the value of the specified gene.
void	setGene(int chromosome, int gene, int value) Sets the value of the specified gene.
void	setIterations(int value) Sets the iterations to use.
void	setMaxTrainTime(int value) Sets the maximum number of seconds to perform training.
void	setNumChrom(int value) Sets the number of chromosomes to use.
void	setSeed(int value) Sets the seed value to use, resets the random number generator.
void	sort() Sorts genes and fitness arrays according to fitness.
void	stop() Stops the execution of the algorithm.
protected Properties	storeSetup(weka.core.Instances data) Generates a Properties file that stores information on the setup of the genetic algorithm.
protected weka.core.Instances	updateHeader(weka.core.Instances data) Creates a new dataset, with the setup as the new relation name.

Methods inherited from class adams.optimise.AbstractOptimiser

checkBest, cleanUp, compareTo, destroy, equals, getOptimisers, optimise, shallowCopy, shallowCopy

Methods inherited from class adams.core.option.AbstractOptionHandler

cleanUpOptions, debug, debug, debugLevelTipText, finishInit, getDebugLevel, getOptionManager, globalInfo, isDebugOn, newOptionManager, setDebugLevel, toCommandLine, toString

Methods inherited from class adams.core.ConsoleObject

getDebugging, getSystemErr, getSystemOut, sizeOf

Methods inherited from class java.lang.Object

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

Field Detail

PROPS_RELATION

public static final String PROPS_RELATION

the key for the relation name in the generated properties file.

See Also:

#storeSetup(Instances,GeneticAlgorithmJob)., Constant Field Values

PROPS_FILTER

public static final String PROPS_FILTER

the key for a filter setup in the setup properties.

See Also:

Constant Field Values

PROPS_MASK

public static final String PROPS_MASK

the key for the mask in the setup properties.

See Also:

Constant Field Values

m_NumGenes

protected int m_NumGenes

number of genes per chromosome. NB: must be initialized by the algorithm!

m_NumChrom

protected int m_NumChrom

number of chromosomes.

m_Genes

protected BitSet[] m_Genes

the genes.

m_Fitness

protected double[] m_Fitness

the fitness of the genes.

m_BestRange

protected Range m_BestRange

stores the best range of attribtues.

m_FavorZeroes

protected boolean m_FavorZeroes

whether to favor 0s instead of 1s.

m_Seed

protected int m_Seed

the seed value.

m_Random

protected Random m_Random

the random number generator.

m_Running

protected boolean m_Running

whether the algorithm is still running.

m_MaxTrainTime

protected int m_MaxTrainTime

the maximum number of seconds to train.

m_TrainStart

protected long m_TrainStart

the time when training commenced.

m_Iterations

protected int m_Iterations

number of iterations.

Constructor Detail

AbstractGeneticAlgorithm

public AbstractGeneticAlgorithm()

Method Detail

getInitialSetups

public abstract Vector<int[]> getInitialSetups()

initialize

protected void initialize()

Initializes the members.

Overrides:

initialize in class AbstractOptimiser

reset

public void reset()

Resets the genetic algorihtm. Derived classes must call this method in set-methods of parameters to assure the invalidation of previously generated data.

Overrides:

reset in class AbstractOptimiser

defineOptions

public void defineOptions()

Adds options to the internal list of options.

Specified by:

defineOptions in interface OptionHandler

Overrides:

defineOptions in class AbstractOptionHandler

setBestRange

protected void setBestRange(Range value)

Sets the best range of attributes.

Parameters:

value - the range

setBestRange

public void setBestRange(String value)

Sets the best range of attributes.

Parameters:

value - the range

getBestRange

public String getBestRange()

Returns the best range of attributes.

Returns:

the range

bestRangeTipText

public String bestRangeTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

setSeed

public void setSeed(int value)

Sets the seed value to use, resets the random number generator.

Parameters:

value - the seed to use

getSeed

public int getSeed()

Returns the current seed value.

Returns:

the seed value

seedTipText

public String seedTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

setFavorZeroes

public void setFavorZeroes(boolean value)

Sets whether 0s are favored over 1s.

Parameters:

value - if true then 0s are favored over 1s

getFavorZeroes

public boolean getFavorZeroes()

Returns whether 0s are favored over 1s.

Returns:

true if 0s are favored over 1s

favorZeroesTipText

public String favorZeroesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

getNumGenes

public int getNumGenes()

Returns the number of genes to use.

Returns:

the number

setNumChrom

public void setNumChrom(int value)

Sets the number of chromosomes to use.

Parameters:

value - the number

getNumChrom

public int getNumChrom()

Returns the number of chromosomes to use.

Returns:

the number

numChromTipText

public String numChromTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

setMaxTrainTime

public void setMaxTrainTime(int value)

Sets the maximum number of seconds to perform training.

Parameters:

value - the number of seconds

getMaxTrainTime

public int getMaxTrainTime()

Returns the maximum number of seconds to perform training.

Returns:

the seed value

maxTrainTimeTipText

public String maxTrainTimeTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

stop

public void stop()

Stops the execution of the algorithm.

isRunning

public boolean isRunning()

Returns whether the algorithm is still running.

Returns:

true if the algorithm is still running

getFitness

public double[] getFitness()

Returns the current fitness values.

Returns:

the fitness values

setIterations

public void setIterations(int value)

Sets the iterations to use.

Parameters:

value - the iterations.

getIterations

public int getIterations()

Gets the iterations use.

Returns:

the iterations

iterationsTipText

public String iterationsTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the GUI or for listing the options.

init

protected void init(int ch,
                    int genes)

Initializes the algorithm.

Parameters:

ch - the chromosome

genes - the number of genes

getGene

public boolean getGene(int chromosome,
                       int gene)

Returns the value of the specified gene.

Parameters:

chromosome - the chromosome index

gene - the gene index

Returns:

true if gene is active

setGene

public void setGene(int chromosome,
                    int gene,
                    int value)

Sets the value of the specified gene.

Parameters:

chromosome - the chromosome index

gene - the gene index

value - the value to set, 0=clear, otherwise activate it

setGene

public void setGene(int chromosome,
                    int gene,
                    boolean value)

Sets the value of the specified gene.

Parameters:

chromosome - the chromosome index

gene - the gene index

value - false=clear, true=activate it

hasMoreZeroes

public boolean hasMoreZeroes(BitSet a,
                             BitSet b)

return if a has more zeroes than b.

Parameters:

a - the first gene

b - the second gene

Returns:

true if a has more zeroes than b

sort

public void sort()

Sorts genes and fitness arrays according to fitness.

doCrossovers

public void doCrossovers()

Performs cross-over.

copyGene

protected void copyGene(int to,
                        int from)

Copies the values of one gene to another.

Parameters:

to - the gene to copy to

from - the gene to copy from

doMutations

public void doMutations()

Performs mutations.

doMutations2

public void doMutations2()

Performs mutations.

calcFitness

public abstract void calcFitness()

Override the following function in sub-classes.

storeSetup

protected Properties storeSetup(weka.core.Instances data)

Generates a Properties file that stores information on the setup of the genetic algorithm. E.g., it backs up the original relation name. The generated properties file will be used as new relation name for the data. Derived classes can add additional parameters to this properties file.

Parameters:

data - the data to create the setup for

job - the associated job

Returns:

the generated setup

See Also:

PROPS_RELATION

updateHeader

protected weka.core.Instances updateHeader(weka.core.Instances data)

Creates a new dataset, with the setup as the new relation name.

Parameters:

data - the data to replace the relation name with the setup

job - the associated job

Returns:

the updated dataset

preRun

protected void preRun()

Further initializations in derived classes.

run

public boolean run()

Runs the genetic algorithm.

Returns:

true if successfully finished, false when interrupted

postRun

protected void postRun()
                throws Exception

Further clean-ups in derived classes.

Throws:

Exception - if something goes wrong

runGeneticAlgorithm

public static void runGeneticAlgorithm(Class env,
                                       Class genetic,
                                       String[] options)

Runs the genetic algorithm with the given options.

Parameters:

env - the environment class to use

genetic - the genetic algorithm to run

options - the options to set

getGeneticAlgorithms

public static String[] getGeneticAlgorithms()

Returns a list with classnames of genetic algorithms.

Returns:

the genetic algorithm classnames

forName

public static AbstractGeneticAlgorithm forName(String classname,
                                               String[] options)

Instantiates the genetic algorithm with the given options.

Parameters:

classname - the classname of the genetic algorithm to instantiate

options - the options for the genetic algorithm

Returns:

the instantiated genetic algorithm or null if an error occurred

forCommandLine

public static AbstractGeneticAlgorithm forCommandLine(String cmdline)

Instantiates the genetic algorithm from the given commandline (i.e., classname and optional options).

Parameters:

cmdline - the classname (and optional options) of the genetic algorithm to instantiate

Returns:

the instantiated genetic algorithm or null if an error occurred