org.vizzini.example.tictactoe
Class NeuralNetworkEvaluator

java.lang.Object
  org.vizzini.game.AbstractEvaluator
      org.vizzini.game.AbstractNeuralNetworkEvaluator
          org.vizzini.game.boardgame.AbstractNeuralGridBoardEvaluator
              org.vizzini.example.tictactoe.NeuralNetworkEvaluator

All Implemented Interfaces:

Serializable, IEvaluator

public class NeuralNetworkEvaluator
extends AbstractNeuralGridBoardEvaluator

Provides a neural network evaluator for tic-tac-toe.

Since:

v0.1

Version:

v0.2

Author:

Jeffrey M. Thompson

See Also:

Serialized Form

Constructor Summary

NeuralNetworkEvaluator()

Method Summary

protected double[]

encode(IGridBoard gridBoard, IAgent agent) Encode the given grid board as appropriate for a neural network input.

Methods inherited from class org.vizzini.game.boardgame.AbstractNeuralGridBoardEvaluator

arrayToString, evaluateTokens

Methods inherited from class org.vizzini.game.AbstractNeuralNetworkEvaluator

configure, getNeuralNetwork, setNeuralNetwork

Methods inherited from class org.vizzini.game.AbstractEvaluator

evaluate, multiplier

Methods inherited from class java.lang.Object

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

Constructor Detail

NeuralNetworkEvaluator

public NeuralNetworkEvaluator()

Method Detail

encode

protected double[] encode(IGridBoard gridBoard,
                          IAgent agent)

Encode the given grid board as appropriate for a neural network input.

The tic-tac-toe board has nine positions. Each position contains an empty, X, or O. The neural network therefore has 18 inputs, 2 for each position.

Specified by:

encode in class AbstractNeuralGridBoardEvaluator

Since:

v0.1