示例#1
0
        static Task <STuple <double, double> > FitnessFunction(INeuralNetwork n1, INeuralNetwork n2)
        {
            #region sequence

            /*
             * for (int n = 0; n < 10; n++)
             * {
             *  Random random = new Random();
             *  var options = Enumerable.Range(0, Amount).ToList();
             *  var test = new double[Amount];
             *  for (int i = 0; i < Amount; i++)
             *  {
             *      int r = random.Next(options.Count);
             *      var next = (double)options[r];
             *      next = Extensions.Map(next, 0, (double)Amount, -1, 1);
             *      test[i] = next;
             *      options.RemoveAt(r);
             *  }
             *  if (test.Length != Amount)
             *  {
             *      Console.WriteLine("wtf!");
             *  }
             *  var result = network.Activate(test);
             *  for (int i = 0; i < Amount; i++)
             *  {
             *      double loss = Math.Abs(result[i] - test[i]);
             *      fitness -= loss;
             *      if (debug && n == 0)
             *          Console.WriteLine($"{test[i]} | {result[i]}");
             *  }
             * }
             */
            #endregion
            #region XOR

            /*
             * var tests = new Dictionary<double[], double>
             * {
             *  {
             *      new double[] { 0, 0},
             *      0
             *  },
             *  {
             *      new double[] { 1, 0},
             *      1
             *  },
             *  {
             *      new double[] { 0, 1},
             *      1
             *  },
             *  {
             *      new double[] { 1, 1},
             *      0
             *  },
             * };
             *
             * foreach(var kv in tests)
             * {
             *  var result = network.Activate(kv.Key);
             *  double difference = result[0] - kv.Value;
             *  difference = -Math.Abs(difference);
             *  fitness += difference;
             *  if (debug)
             *  {
             *      Console.WriteLine($"{kv.Key[0]} {kv.Key[1]} | {result[0]}");
             *  }
             * }
             */
            #endregion XOR
            #region linear

            /*
             * for (int i = 0; i < 10; i++)
             * {
             *  double a, b, x;
             *  lock (random)
             *  {
             *      a = Math.Round(random.NextDouble() / 2, 2);
             *      b = Math.Round(random.NextDouble() / 2, 2);
             *      x = Math.Round(random.NextDouble(), 2);
             *  }
             *  var res = network.Activate(new double[] { a, b, x });
             *  var correct = (a * x + b);
             *
             *  fitness -= Math.Abs(res[0] - correct);
             *
             *  if (debug)
             *  {
             *      Console.WriteLine($"{a} * {x} + {b} = {correct} | {res[0]}");
             *  }
             * }
             */
            #endregion
            #region TicTacToe
            Game           game      = new Game();
            Game.GameState gameState = Game.GameState.Undetermined;

            bool n1turn = true;

            while (gameState == Game.GameState.Undetermined)
            {
                var board = BoardToArray(game.GetBoard());
                var input = new double[10];
                input[0] = n1turn ? 1 : -1;
                for (int i = 0; i < board.Length; i++)
                {
                    input[i + 1] = board[i];
                }
                INeuralNetwork network = n1turn ? n1 : n2;
                var            output  = network.Activate(input);

                var        newGameState = Game.GameState.Error;
                List <int> chosenTiles  = new List <int>();
                while (newGameState == Game.GameState.Error)
                {
                    int    tile    = -1;
                    double highest = double.MinValue;
                    for (int i = 0; i < output.Length; i++)
                    {
                        if (output[i] > highest && !chosenTiles.Contains(i))
                        {
                            highest = output[i];
                            tile    = i;
                        }
                    }
                    chosenTiles.Add(tile);
                    tile++;

                    newGameState = game.NewMove((Game.TileNumber)tile);
                }

                n1turn    = !n1turn;
                gameState = newGameState;
            }

            (double, double)fitness = (0, 0);
            if (gameState == Game.GameState.Win)
            {
                if (n1turn)
                {
                    fitness = (-1, 1);
                }
                else
                {
                    fitness = (1, -1);
                }
            }

            #endregion

            var tcs = new TaskCompletionSource <STuple <double, double> >();
            tcs.SetResult(fitness);
            return(tcs.Task);
        }
示例#2
0
        static void HumanTest(INeuralNetwork enemy)
        {
            Console.Clear();
            Game game = new Game();

            Game.GameState gameState = Game.GameState.Undetermined;

            bool playerturn = true;

            while (gameState == Game.GameState.Undetermined)
            {
                if (!playerturn)
                {
                    var board = BoardToArray(game.GetBoard());
                    var input = new double[10];
                    input[0] = -1;
                    for (int i = 0; i < board.Length; i++)
                    {
                        input[i + 1] = board[i];
                    }
                    var output = enemy.Activate(input);

                    var        newGameState = Game.GameState.Error;
                    List <int> chosenTiles  = new List <int>();
                    while (newGameState == Game.GameState.Error)
                    {
                        int    tile    = -1;
                        double highest = double.MinValue;
                        for (int i = 0; i < output.Length; i++)
                        {
                            if (output[i] > highest && !chosenTiles.Contains(i))
                            {
                                highest = output[i];
                                tile    = i;
                            }
                        }
                        chosenTiles.Add(tile);
                        tile++;

                        newGameState = game.NewMove((Game.TileNumber)tile);
                    }
                    gameState = newGameState;
                }
                else
                {
                    var newGameState = Game.GameState.Error;
                    while (newGameState == Game.GameState.Error)
                    {
                        int tile = -1;
                        int.TryParse(Console.ReadLine(), out tile);
                        newGameState = game.NewMove((Game.TileNumber)tile);
                    }
                    gameState = newGameState;
                }
                playerturn = !playerturn;
                Console.WriteLine();
                VisualizeBoard(game.GetBoard());
                Console.WriteLine();
            }

            if (gameState == Game.GameState.Win)
            {
                string winner = playerturn ? "Bot" : "Player";
                Console.WriteLine();
                Console.WriteLine(winner + " wins!");
            }
            else if (gameState == Game.GameState.Tie)
            {
                Console.WriteLine();
                Console.WriteLine("Tie!");
            }
            Console.ReadLine();
        }