Package net.bmahe.genetics4j.neat

Class RecurrentNetwork

java.lang.Object

net.bmahe.genetics4j.neat.RecurrentNetwork

public class RecurrentNetwork
extends Object

Implements a recurrent neural network evaluator for NEAT chromosomes.

Unlike FeedForwardNetwork, this implementation can execute arbitrary directed graphs that include recurrent (cyclic) connections. Activations are propagated iteratively until they converge or a maximum number of iterations is reached, making it suitable for tasks that rely on short-term memory or feedback loops.

Field Summary

Fields

Modifier and Type	Field	Description
private final Function<Float,Float>	activationFunction
private final Set<Integer>	allNodeIndices
private final Map<Integer,Set<Connection>>	backwardConnections
private final List<Connection>	connections
private final float	convergenceThreshold
static final float	DEFAULT_CONVERGENCE_THRESHOLD	Default convergence threshold applied to node deltas between iterations.
static final float	DEFAULT_INITIAL_STATE_VALUE	Default value assigned to non-input nodes before the first iteration.
static final int	DEFAULT_MAX_ITERATIONS	Default maximum number of recurrent iterations.
private final List<Integer>	evaluatedNodeIndices
private final float	initialStateValue
private final Set<Integer>	inputNodeIndices
static final org.apache.logging.log4j.Logger	logger
private final int	maxIterations
private final Map<Integer,Float>	nodeState
private final Set<Integer>	outputNodeIndices

Constructor Summary

Constructors

Constructor	Description
RecurrentNetwork(Set<Integer> _inputNodeIndices, Set<Integer> _outputNodeIndices, List<Connection> _connections, Function<Float,Float> _activationFunction)
RecurrentNetwork(Set<Integer> _inputNodeIndices, Set<Integer> _outputNodeIndices, List<Connection> _connections, Function<Float,Float> _activationFunction, int _maxIterations, float _convergenceThreshold, float _initialStateValue)

Method Summary

Modifier and Type	Method	Description
private Map<Integer,Float>	buildOutputValues(Map<Integer,Float> nodeValues)
Map<Integer,Float>	compute(Map<Integer,Float> inputValues)
void	resetState()
Map<Integer,Float>	step(Map<Integer,Float> inputValues)

Methods inherited from class java.lang.Object

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

Field Details

logger

public static final org.apache.logging.log4j.Logger logger

DEFAULT_MAX_ITERATIONS

public static final int DEFAULT_MAX_ITERATIONS

Default maximum number of recurrent iterations.

See Also:

• Constant Field Values

DEFAULT_CONVERGENCE_THRESHOLD

public static final float DEFAULT_CONVERGENCE_THRESHOLD

Default convergence threshold applied to node deltas between iterations.

See Also:

• Constant Field Values

DEFAULT_INITIAL_STATE_VALUE

public static final float DEFAULT_INITIAL_STATE_VALUE

Default value assigned to non-input nodes before the first iteration.

See Also:

• Constant Field Values

inputNodeIndices

private final Set<Integer> inputNodeIndices

outputNodeIndices

private final Set<Integer> outputNodeIndices

connections

private final List<Connection> connections

backwardConnections

private final Map<Integer,Set<Connection>> backwardConnections

evaluatedNodeIndices

private final List<Integer> evaluatedNodeIndices

allNodeIndices

private final Set<Integer> allNodeIndices

activationFunction

private final Function<Float,Float> activationFunction

maxIterations

private final int maxIterations

convergenceThreshold

private final float convergenceThreshold

initialStateValue

private final float initialStateValue

nodeState

private final Map<Integer,Float> nodeState

Constructor Details

RecurrentNetwork

public RecurrentNetwork(Set<Integer> _inputNodeIndices,
 Set<Integer> _outputNodeIndices,
 List<Connection> _connections,
 Function<Float,Float> _activationFunction)

RecurrentNetwork

public RecurrentNetwork(Set<Integer> _inputNodeIndices,
 Set<Integer> _outputNodeIndices,
 List<Connection> _connections,
 Function<Float,Float> _activationFunction,
 int _maxIterations,
 float _convergenceThreshold,
 float _initialStateValue)

Method Details

compute

public Map<Integer,Float> compute(Map<Integer,Float> inputValues)

step

public Map<Integer,Float> step(Map<Integer,Float> inputValues)

resetState

public void resetState()

buildOutputValues

private Map<Integer,Float> buildOutputValues(Map<Integer,Float> nodeValues)