Class XCSConstants

java.lang.Object
  |
  +--XCSConstants

All Implemented Interfaces:

java.io.Serializable

public class XCSConstants

extends java.lang.Object

implements java.io.Serializable

This class provides all relevant learning parameters for the XCS as well as other experimental settings and flags. Most parameter-names are chosen similar to the 'An Algorithmic Description of XCS' ( Butz&Wilson, IlliGAL report 2000017).

Since:

JDK 1.3

See Also:

Serialized Form

Field Summary

private static long	_A Constant for the random number generator (default = 16807).
private static long	_M Constant for the random number generator (modulus of PMMLCG = 2^31 -1).
private static long	_Q Constant for the random number generator (=_M/_A).
private static long	_R Constant for the random number generator (=_M mod _A).
static double	alpha The fall of rate in the fitness evaluation.
static double	beta The learning rate for updating fitness, prediction, prediction error, and action set size estimate in XCS's classifiers.
static double	delta The fraction of the mean fitness of the population below which the fitness of a classifier may be considered in its vote for deletion.
static boolean	doActionSetSubsumption Specifies if action set subsumption should be executed.
static boolean	doGASubsumption Specifies if GA subsumption should be executed.
static char	dontCare The don't care symbol (normally '#')
static double	epsilon_0 The error threshold under which the accuracy of a classifier is set to one.
static double	fitnessIni The initial prediction value when generating a new classifier (e.g in covering).
static double	fitnessReduction The reduction of the fitness when generating an offspring classifier.
static double	gamma The discount rate in multi-step problems.
static int	maxPopSize Specifies the maximal number of micro-classifiers in the population.
static double	nu Specifies the exponent in the power function for the fitness evaluation.
static double	P_dontcare The probability of using a don't care symbol in an allele when covering.
static double	pM The probability of mutating one allele and the action in an offspring classifier.
static double	predictionErrorIni The initial prediction error value when generating a new classifier (e.g in covering).
static double	predictionErrorReduction The reduction of the prediction error when generating an offspring classifier.
static double	predictionIni The initial prediction value when generating a new classifier (e.g in covering).
static double	pX The probability of applying crossover in an offspring classifier.
private static long	seed The initialization of the pseudo random generator.
static int	teletransportation The maximal number of steps executed in one trial in a multi-step problem.
static int	theta_del Specified the threshold over which the fitness of a classifier may be considered in its deletion probability.
static double	theta_GA The threshold for the GA application in an action set.
static int	theta_sub The experience of a classifier required to be a subsumer.

Constructor Summary

XCSConstants()
The default constructor.

Method Summary

static float	drand() Returns a random number in between zero and one.
static void	setSeed(long s) Sets a random seed in order to randomize the pseudo random generator.

Methods inherited from class java.lang.Object

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

Field Detail

maxPopSize

public static final int maxPopSize

Specifies the maximal number of micro-classifiers in the population.

See Also:

Constant Field Values

alpha

public static final double alpha

The fall of rate in the fitness evaluation.

See Also:

Constant Field Values

beta

public static final double beta

The learning rate for updating fitness, prediction, prediction error, and action set size estimate in XCS's classifiers.

See Also:

Constant Field Values

gamma

public static final double gamma

The discount rate in multi-step problems.

See Also:

Constant Field Values

delta

public static final double delta

The fraction of the mean fitness of the population below which the fitness of a classifier may be considered in its vote for deletion.

See Also:

Constant Field Values

nu

public static final double nu

Specifies the exponent in the power function for the fitness evaluation.

See Also:

Constant Field Values

theta_GA

public static final double theta_GA

The threshold for the GA application in an action set.

See Also:

Constant Field Values

epsilon_0

public static final double epsilon_0

The error threshold under which the accuracy of a classifier is set to one.

See Also:

Constant Field Values

theta_del

public static final int theta_del

Specified the threshold over which the fitness of a classifier may be considered in its deletion probability.

See Also:

Constant Field Values

pX

public static final double pX

The probability of applying crossover in an offspring classifier.

See Also:

Constant Field Values

pM

public static final double pM

The probability of mutating one allele and the action in an offspring classifier.

See Also:

Constant Field Values

P_dontcare

public static final double P_dontcare

The probability of using a don't care symbol in an allele when covering.

See Also:

Constant Field Values

predictionErrorReduction

public static final double predictionErrorReduction

The reduction of the prediction error when generating an offspring classifier.

See Also:

Constant Field Values

fitnessReduction

public static final double fitnessReduction

The reduction of the fitness when generating an offspring classifier.

See Also:

Constant Field Values

theta_sub

public static final int theta_sub

The experience of a classifier required to be a subsumer.

See Also:

Constant Field Values

teletransportation

public static final int teletransportation

The maximal number of steps executed in one trial in a multi-step problem.

See Also:

Constant Field Values

doGASubsumption

public static final boolean doGASubsumption

Specifies if GA subsumption should be executed.

See Also:

Constant Field Values

doActionSetSubsumption

public static final boolean doActionSetSubsumption

Specifies if action set subsumption should be executed.

See Also:

Constant Field Values

predictionIni

public static final double predictionIni

The initial prediction value when generating a new classifier (e.g in covering).

See Also:

Constant Field Values

predictionErrorIni

public static final double predictionErrorIni

The initial prediction error value when generating a new classifier (e.g in covering).

See Also:

Constant Field Values

fitnessIni

public static final double fitnessIni

The initial prediction value when generating a new classifier (e.g in covering).

See Also:

Constant Field Values

dontCare

public static final char dontCare

The don't care symbol (normally '#')

See Also:

Constant Field Values

seed

private static long seed

The initialization of the pseudo random generator. Must be at least one and smaller than _M.

_M

private static final long _M

Constant for the random number generator (modulus of PMMLCG = 2^31 -1).

See Also:

Constant Field Values

_A

private static final long _A

Constant for the random number generator (default = 16807).

See Also:

Constant Field Values

_Q

private static final long _Q

Constant for the random number generator (=_M/_A).

See Also:

Constant Field Values

_R

private static final long _R

Constant for the random number generator (=_M mod _A).

See Also:

Constant Field Values

Constructor Detail

XCSConstants

public XCSConstants()

The default constructor.

Method Detail

setSeed

public static void setSeed(long s)

Sets a random seed in order to randomize the pseudo random generator.

drand

public static float drand()

Returns a random number in between zero and one.