org.encog.neural.networks.training.genetic
Class NeuralChromosome

java.lang.Object
  org.encog.solve.genetic.Chromosome<java.lang.Double>
      org.encog.neural.networks.training.genetic.NeuralChromosome

All Implemented Interfaces:

java.lang.Comparable<Chromosome<java.lang.Double>>

public class NeuralChromosome
extends Chromosome<java.lang.Double>

Implements a chromosome that allows a feedforward neural network to be trained using a genetic algorithm. The chromosome for a feed forward neural network is the weight and threshold matrix. This class is abstract. If you wish to train the neural network using training sets, you should use the TrainingSetNeuralChromosome class. If you wish to use a score function to train the neural network, then implement a subclass of this one that properly calculates the score. The generic type GA_TYPE specifies the GeneticAlgorithm derived class that implements the genetic algorithm that this class is to be used with.

Field Summary

static double

DISTORT_FACTOR The amount of distortion to perform a mutation.

Constructor Summary

NeuralChromosome(NeuralGeneticAlgorithm genetic, BasicNetwork network)
Construct a neural chromosome.

Method Summary

void	calculateScore() Calculate the score for this chromosome.
BasicNetwork	getNetwork()
void	initGenes(int length) Init the genes array.
void	mutate() Mutate this chromosome randomly.
void	setGenes(java.lang.Double[] list) Set all genes.
void	setNetwork(BasicNetwork network)
void	updateGenes() Copy the network to the genes.
void	updateNetwork() Copy the genes to the network.

Methods inherited from class org.encog.solve.genetic.Chromosome

compareTo, getGene, getGenes, getGeneticAlgorithm, getScore, mate, setGene, setGenesDirect, setGeneticAlgorithm, setScore, toString

Methods inherited from class java.lang.Object

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

Field Detail

DISTORT_FACTOR

public static final double DISTORT_FACTOR

The amount of distortion to perform a mutation.

See Also:

Constant Field Values

Constructor Detail

NeuralChromosome

public NeuralChromosome(NeuralGeneticAlgorithm genetic,
                        BasicNetwork network)

Construct a neural chromosome.

Parameters:

genetic - The genetic algorithm that uses this chromosome.

network - The network that this chromosome is based on.

Method Detail

getNetwork

public BasicNetwork getNetwork()

Returns:

the network

initGenes

public void initGenes(int length)

Init the genes array.

Parameters:

length - The length to create.

mutate

public void mutate()

Mutate this chromosome randomly.

Specified by:

mutate in class Chromosome<java.lang.Double>

setGenes

public void setGenes(java.lang.Double[] list)

Set all genes.

Overrides:

setGenes in class Chromosome<java.lang.Double>

Parameters:

list - A list of genes.

setNetwork

public void setNetwork(BasicNetwork network)

Parameters:

network - the network to set

updateGenes

public void updateGenes()

Copy the network to the genes.

updateNetwork

public void updateNetwork()

Copy the genes to the network.

calculateScore

public void calculateScore()

Calculate the score for this chromosome.

Specified by:

calculateScore in class Chromosome<java.lang.Double>