// CONSTRUCTEUR
/// <summary>
/// Creates an instance of the DataCollection class
/// </summary>
/// <param name="a_parser">Parser for getting examples from a given data source</param>
/// <param name="a_inputsNumber">Number of inputs contained in an example</param>
/// <param name="a_outputsNumber">Number of outputs contained in an example</param>
/// <param name="a_learningRatio">Ratio (between 0 and 1) used for examples used for the neural network training, others examples are used for overlearning detection</param>
public DataCollection(IExempleDataParserRepository a_parser, int a_inputsNumber, int a_outputsNumber, double a_learningRatio)
{
// Validation des arguments
if (a_learningRatio < 0.0 || a_learningRatio > 1.0)
{
throw new ArgumentException("a_trainingRatio argument should be between 0.0 and 1.0 (0.8 value is recommended).");
}
List <DataPoint> dataFromParser = a_parser.GetData();
Random generator = new Random();
int trainingExempleNumber = (int)(dataFromParser.Count * a_learningRatio);
int generalisationExempleNumber = dataFromParser.Count - trainingExempleNumber;
TrainingPoints = new DataPoint[trainingExempleNumber];
GeneralisationPoints = new DataPoint[generalisationExempleNumber];
if (trainingExempleNumber > 0)
{
int effectiveTrainingExempleNumber = 0;
// Remplissage du tableau trainingPoints avec des données des exemples de façon aléatoire
while (trainingExempleNumber > effectiveTrainingExempleNumber)
{
int randomIndex = generator.Next(dataFromParser.Count - 1);
TrainingPoints[effectiveTrainingExempleNumber] = dataFromParser[randomIndex];
dataFromParser.RemoveAt(randomIndex);
effectiveTrainingExempleNumber++;
}
for (int i = 0; i < generalisationExempleNumber; i++)
{
GeneralisationPoints[i] = dataFromParser[i];
}
}
else
{
TrainingPoints = null;
GeneralisationPoints = null;
}
}
// CONSTRUCTEUR
public NeuralSystem(int a_inputsNumber, int a_hiddenNeuronsNumber, int a_outputNeuronsNumber, IExempleDataParserRepository a_parser, double a_learningRatio, IUserInterface a_gui, bool a_regressionTask)
{
data = new DataCollection(a_parser, a_inputsNumber, a_outputNeuronsNumber, a_learningRatio);
network = new NeuralNetwork(a_inputsNumber, a_hiddenNeuronsNumber, a_outputNeuronsNumber);
gui = a_gui;
regressionTask = a_regressionTask;
LearningRate = 0.3;
MaxError = 0.005;
MaxIterations = 10001;
}
