net.sourceforge.cilib.games.agent.state.evaluation
Class NeuralStateEvaluator

java.lang.Object
  net.sourceforge.cilib.games.agent.Agent
      net.sourceforge.cilib.games.agent.NeuralAgent
          net.sourceforge.cilib.games.agent.state.evaluation.NeuralStateEvaluator

All Implemented Interfaces:

Serializable, StateEvaluator, Cloneable

public class NeuralStateEvaluator
extends NeuralAgent
implements StateEvaluator

Author:

leo This is a StateEvaluator that uses a Neural Network to evaluate a given game state. This class inherets from the NeuralAgent class becuase it's functionality is very similar.

See Also:

Serialized Form

Field Summary

Fields inherited from class net.sourceforge.cilib.games.agent.NeuralAgent

hiddenNodesCount, neuralNetworkTopology, outputInterpretationStrategy, stateInputStrategy

Fields inherited from class net.sourceforge.cilib.games.agent.Agent

agentScore, agentToken, playerID

Constructor Summary

NeuralStateEvaluator()

NeuralStateEvaluator(NeuralStateEvaluator other)
Copy Constructor

Method Summary

double	evaluateState(Game<GameState> state, int decisionPlayerID) Evaluate a given GameState for a given player ID.
NeuralStateEvaluator	getClone() Create a cloned copy of the current object and return it.
DomainRegistry	getEvaluatorDomain() If the evaluator can be optimized then the evaluator data should be in a specific domain, this function should return that domain
void	initializeEvaluator(Type evaluatorData) Some state evaluators can be optimised, and therefore the evaluator should be initialized with evaluator specific data.
void	move(Game game) Alter the current game by making a decision.
void	setHiddenNodes(int count)
void	setOutputInterpretationStrategy(NeuralOutputInterpretationStrategy outputStrategy) The NeuralOutputInterpretationStrategy should not be set since this class funtions only as a state evaluator.
void	setStateInputStrategy(NeuralStateInputStrategy inputStrategy)

Methods inherited from class net.sourceforge.cilib.games.agent.NeuralAgent

getAgentDomain, getHiddenNodesCount, getNeuralNetworkTopology, getOutputInterpretationStrategy, getScaledInput, getStateInputStrategy, initializeAgent, setNeuralNetworkTopology, setWeights

Methods inherited from class net.sourceforge.cilib.games.agent.Agent

getAgentScore, getAgentToken, getPlayerID, setAgentScore, setAgentToken, setAgentToken, setPlayerID

Methods inherited from class java.lang.Object

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

Constructor Detail

NeuralStateEvaluator

public NeuralStateEvaluator()

NeuralStateEvaluator

public NeuralStateEvaluator(NeuralStateEvaluator other)

Copy Constructor

Parameters:

other -

Method Detail

evaluateState

public double evaluateState(Game<GameState> state,
                            int decisionPlayerID)

Evaluate a given GameState for a given player ID.

Specified by:

evaluateState in interface StateEvaluator

Parameters:

state - the Game object that is at a specific state that needs to be evaluated

decisionPlayerID - The player that has to make a decision

Returns:

the rating of the state.

getEvaluatorDomain

public DomainRegistry getEvaluatorDomain()

If the evaluator can be optimized then the evaluator data should be in a specific domain, this function should return that domain

Specified by:

getEvaluatorDomain in interface StateEvaluator

Returns:

The domain of the evaluator

initializeEvaluator

public void initializeEvaluator(Type evaluatorData)

Some state evaluators can be optimised, and therefore the evaluator should be initialized with evaluator specific data.

Specified by:

initializeEvaluator in interface StateEvaluator

Parameters:

evaluatorData - the data that determines how the evaluator should funciton

getClone

public NeuralStateEvaluator getClone()

Create a cloned copy of the current object and return it. In general the created copy will be a deep copy of the provided instance. As a result this operation an be quite expensive if used incorrectly.

Specified by:

getClone in interface Cloneable

Overrides:

getClone in class NeuralAgent

Returns:

An exact clone of the current object instance.

See Also:

Object.clone()

setHiddenNodes

public void setHiddenNodes(int count)

Overrides:

setHiddenNodes in class NeuralAgent

setStateInputStrategy

public void setStateInputStrategy(NeuralStateInputStrategy inputStrategy)

Overrides:

setStateInputStrategy in class NeuralAgent

setOutputInterpretationStrategy

public void setOutputInterpretationStrategy(NeuralOutputInterpretationStrategy outputStrategy)

The NeuralOutputInterpretationStrategy should not be set since this class funtions only as a state evaluator. A single output Neuron is used to rank the given state.

Overrides:

setOutputInterpretationStrategy in class NeuralAgent

move

public void move(Game game)

Alter the current game by making a decision.

Overrides:

move in class NeuralAgent

Parameters:

game - the game state