//tinh mse cho moi ca the
//private double CalculateMSE(Individual_NSGA individual, double[][] inputData)
//{
// // Chỗ này chính là chỗ cần dùng với RBF đây
// RadialNetwork rn = new RadialNetwork(m_RadialNetwork);
// double fitness = 0.0;
// rn.SetWeights(individual.values);
// fitness = rn.Accuracy(inputData);
// return fitness;
//}
//tinh mse cho moi ca the
private double CalculateMSE(Individual_NSGA individual, double[][] inputData)
{
// Chỗ này chính là chỗ cần dùng với RBF đây
RadialNetwork rn = new RadialNetwork(m_RadialNetwork);
double fitness = 0.0;
rn.SetWeights(individual.values);
fitness = rn.getMSEOf(inputData);
return(fitness);
}
private double CalculateFitnessOf(Individual_GA individual, double[][] inputData)
{
// Chỗ này chính là chỗ cần dùng với RBF đây
RadialNetwork rn = new RadialNetwork(m_RadialNetwork);
double fitness = 0.0;
rn.SetWeights(individual.values);
fitness = rn.Accuracy(inputData);
individual.Fitness = fitness;
return(fitness);
}
public NSGAII(int m_Population_size, int m_IndividualLength, RadialNetwork RadialNetwork, double m_MaxEvaluations, double[][] inputData, int numberOfObjectives)
{
this.m_Population_size = m_Population_size;
this.m_IndividualLength = m_IndividualLength;
this.m_Random = new Random();
this.Population = new Population_NSGA(m_Population_size);
this.m_RadialNetwork = RadialNetwork;
this.m_MaxEvaluations = m_MaxEvaluations;
this.inputData = inputData;
mutationProb = 1.0 / m_IndividualLength;
this.Population.Population_init(m_IndividualLength, numberOfObjectives);
}
} // ctor
public RadialNetwork(RadialNetwork newOne)
{
rnd = new Random(0);
this.numInput = newOne.NumInput;
this.numHidden = newOne.NumHidden;
this.numOutput = newOne.NumOutput;
this.inputs = newOne.Inputs;
this.centroids = newOne.Centroids;
this.widths = newOne.Widths;
this.hoWeights = newOne.HoWeights;
this.oBiases = newOne.OBiases;
this.outputs = newOne.Outputs;
} // ctor
private void btnStart_Click(object sender, EventArgs e)
{
//Get input from Form
//Get testing rate
if (!StringIsNull(txtTestingRate.Text))
{
testingRate = double.Parse(txtTestingRate.Text);
}
//Get window size
if (!StringIsNull(txtWindowSize.Text))
{
windowSize = int.Parse(txtWindowSize.Text);
}
//Step1: Get inputTrain, idealTrain, inputTest, idealTest
//GetTrainTest(data, out train, out test, testingRate);
//GetInputIdeal(train, out inputTrain, out idealTrain, windowSize);
//GetInputIdeal(train, out inputTest, out idealTest, windowSize);
GetAllData(data, out allData, windowSize);
double[][] trainData = null;
double[][] testData = null;
int seed = 8; // gives a good demo
GetTrainTest(allData, seed, out trainData, out testData, testingRate);
Console.WriteLine("\nCreating a 4-5-3 radial basis function network");
int numInput = windowSize;
int numHidden = 2 * numInput;
int numOutput = 1;
RadialNetwork rn = new RadialNetwork(numInput, numHidden, numOutput);
Console.WriteLine("\nBeginning RBF training\n");
int maxIterations = 50; // max for GA
int maxEvaluations = 1000; // max for NSGAII
//double[] bestWeights = rn.TrainWithPSO(trainData, maxIterations);
//double[] bestWeights = rn.TrainWithGA(trainData, maxIterations);
Population_NSGA bestPopulation = rn.TrainWithNSGAII(trainData, maxEvaluations);
Console.WriteLine("\nEvaluating result RBF classification accuracy on the test data");
//rn.SetWeights(bestWeights);
double acc = rn.Accuracy(testData);
//Set Textbox
SetText(txtTestErr, acc.ToString("F5"));
Console.WriteLine("Complete Program!");
}
public GAs(int m_Population_size, int m_IndividualLength, RadialNetwork RadialNetwork, double m_MaxIterations, double[][] inputData)
{
this.m_Population_size = m_Population_size;
this.m_IndividualLength = m_IndividualLength;
this.m_Random = new Random();
this.Population = new Population_GA(m_Population_size);
this.m_RadialNetwork = RadialNetwork;
this.m_MaxIterations = m_MaxIterations;
this.Population.Population_init(m_IndividualLength);
// set fitness
for (int i = 0; i < m_Population_size; i++)
{
CalculateFitnessOf(Population.Individuals[i], inputData);
}
}
