public void Configure(IEnumerable<Game> games, IEnumerable<Player> players)
{
m_Players = players.ToArray();
m_GameCount = games.Count();
m_PlayerCount = m_Players.Length;
Random rnd = new Random();
BuildGameData(games.ToArray(), m_Players);
int numInput = m_PlayerCount;
int numHidden = 6;
int numOutput = 1;
int numWeights = (numInput * numHidden) + (numHidden * numOutput) + (numHidden + numOutput);
Console.WriteLine("Creating a " + numInput + "-input, " + numHidden + "-hidden, " + numOutput + "-output neural network");
Console.WriteLine("Using hard-coded tanh function for hidden layer activation");
Console.WriteLine("Using hard-coded log-sigmoid function for output layer activation");
m_Bnn = new BackPropNeuralNet(numInput, numHidden, numOutput);
Console.WriteLine("\nGenerating random initial weights and bias values");
double[] initWeights = new double[numWeights];
for (int i = 0; i < initWeights.Length; ++i)
initWeights[i] = (rnd.NextDouble() - 0.5d) * 0.1d;
//Console.WriteLine("\nInitial weights and biases are:");
//Helpers.ShowVector(initWeights, 3, 8, true);
Console.WriteLine("Loading neural network initial weights and biases into neural network");
m_Bnn.SetWeights(initWeights);
double learnRate = 0.03; // learning rate - controls the maginitude of the increase in the change in weights.
double momentum = learnRate / 10d; // momentum - to discourage oscillation.
Console.WriteLine("Setting learning rate = " + learnRate.ToString("F2") + " and momentum = " + momentum.ToString("F2"));
int maxEpochs = 8000000;
double errorThresh = 0.01;
Console.WriteLine("\nSetting max epochs = " + maxEpochs + " and error threshold = " + errorThresh.ToString("F6"));
int epoch = 0;
double error = double.MaxValue;
Console.WriteLine("\nBeginning training using back-propagation\n");
while (epoch < maxEpochs) // train
{
int gameNum = rnd.Next(m_GameCount); //leave 10 games for prediction
double[] whoPlayed = GetWhoPlayed(gameNum);
double result = GetGameResult(gameNum);
double[] predictedResult = m_Bnn.ComputeOutputs(whoPlayed);
m_Bnn.UpdateWeights(new[] { result }, learnRate, momentum);
++epoch;
if (epoch % 20000 == 0)
{
error = GetTotalError(m_Bnn, true);
if (error < errorThresh)
{
Console.WriteLine("Found weights and bias values that meet the error criterion at epoch " + epoch);
break;
}
Console.WriteLine("epoch = " + epoch);
Console.WriteLine("error = " + error);
}
} // train loop
double[] finalWeights = m_Bnn.GetWeights();
Console.WriteLine("");
//Console.WriteLine("Final neural network weights and bias values are:");
//Helpers.ShowVector(finalWeights, 5, 8, true);
}
private double GetTotalError(BackPropNeuralNet bnn, bool print)
{
double totalError = 0d, totalErrorOnTestData = 0d;
int score = 0, outof = 0;
for (int i = 0; i < m_GameCount; i++)
{
double[] whoPlayed = GetWhoPlayed(i);
double predictedResult = CsvParser.RevConv(bnn.ComputeOutputs(whoPlayed)[0]);
double actualResult = CsvParser.RevConv(GetGameResult(i));
totalError += (predictedResult - actualResult) * (predictedResult - actualResult);
}
if (print)
{
Console.WriteLine("MSE is: {0:0.000}", totalError / (double)m_GameCount);
}
return totalError;
}
