org.encog.neural.neat.training

Class NEATInnovationList

java.lang.Object

org.encog.neural.neat.training.NEATInnovationList

All Implemented Interfaces:

java.io.Serializable

public class NEATInnovationList
extends java.lang.Object
implements java.io.Serializable

Implements a NEAT innovation list. NeuroEvolution of Augmenting Topologies (NEAT) is a genetic algorithm for the generation of evolving artificial neural networks. It was developed by Ken Stanley while at The University of Texas at Austin. ----------------------------------------------------------------------------- http://www.cs.ucf.edu/~kstanley/ Encog's NEAT implementation was drawn from the following three Journal Articles. For more complete BibTeX sources, see NEATNetwork.java. Evolving Neural Networks Through Augmenting Topologies Generating Large-Scale Neural Networks Through Discovering Geometric Regularities Automatic feature selection in neuroevolution

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
NEATInnovationList() The default constructor, used mainly for persistance.
NEATInnovationList(NEATPopulation population) Construct an innovation list, that includes the initial innovations.

Method Summary

Modifier and Type	Method and Description
NEATInnovation	findInnovation(long neuronID) Find an innovation for a single neuron.
NEATInnovation	findInnovation(long fromID, long toID) Find an innovation for a new link added between two existing neurons.
NEATInnovation	findInnovationSplit(long fromID, long toID) Find an innovation for a hidden neuron that split a existing link.
java.util.Map<java.lang.String,NEATInnovation>	getInnovations()
static java.lang.String	produceKeyLink(long fromID, long toID) Produce a key for a link.
static java.lang.String	produceKeyNeuron(long id) Produce an innovation key for a neuron.
static java.lang.String	produceKeyNeuronSplit(long fromID, long toID) Produce a key for a split neuron.
void	setPopulation(NEATPopulation population) Set the population that this genome belongs to.

Methods inherited from class java.lang.Object

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

Constructor Detail

NEATInnovationList

public NEATInnovationList()

The default constructor, used mainly for persistance.

NEATInnovationList

public NEATInnovationList(NEATPopulation population)

Construct an innovation list, that includes the initial innovations.

Parameters:

population - The population to base this innovation list on.

Method Detail

produceKeyNeuron

public static java.lang.String produceKeyNeuron(long id)

Produce an innovation key for a neuron.

Parameters:

id - The neuron id.

Returns:

The newly created key.

produceKeyNeuronSplit

public static java.lang.String produceKeyNeuronSplit(long fromID,
                                                     long toID)

Produce a key for a split neuron.

Parameters:

fromID - Thf from id.

toID - The to id.

Returns:

The key.

produceKeyLink

public static java.lang.String produceKeyLink(long fromID,
                                              long toID)

Produce a key for a link.

Parameters:

fromID - The from id.

toID - The to id.

Returns:

The key for the link.

findInnovationSplit

public NEATInnovation findInnovationSplit(long fromID,
                                          long toID)

Find an innovation for a hidden neuron that split a existing link. This is the means by which hidden neurons are introduced in NEAT.

Parameters:

fromID - The source neuron ID in the link.

toID - The target neuron ID in the link.

Returns:

The newly created innovation, or the one that matched the search.

findInnovation

public NEATInnovation findInnovation(long neuronID)

Find an innovation for a single neuron. Single neurons were created without producing a split. This means, the only single neurons are the input, bias and output neurons.

Parameters:

neuronID - The neuron ID to find.

Returns:

The newly created innovation, or the one that matched the search.

findInnovation

public NEATInnovation findInnovation(long fromID,
                                     long toID)

Find an innovation for a new link added between two existing neurons.

Parameters:

fromID - The source neuron ID in the link.

toID - The target neuron ID in the link.

Returns:

The newly created innovation, or the one that matched the search.

setPopulation

public void setPopulation(NEATPopulation population)

Set the population that this genome belongs to.

Parameters:

population - The population.

getInnovations

public java.util.Map<java.lang.String,NEATInnovation> getInnovations()

Returns:

A list of innovations.