public RadialBasisFunctionNetwork(int _inputCount, int _outputCount, int _rbfCount, List <TeachedRBF> _teachedRBFs, List <OutputLayerWeights> _outputLayerWeights, char[] _classesNames)
{
inputCount = _inputCount;
outputCount = _outputCount;
rbfCount = _rbfCount;
rbf_teached = new RBF_teachInfo[rbfCount];
outputLayer_weights = new double[rbfCount, outputCount];
int i = 0;
foreach (TeachedRBF teachedRBF in _teachedRBFs)
{
RBF_teachInfo rbf = new RBF_teachInfo();
rbf.standardDeviation = teachedRBF.StandartDeviation;
rbf.expectations = Serializer.DeserializeFromString(teachedRBF.Expectation);
rbf_teached[i++] = rbf;
}
i = 0;
outputLayer_weights = new double[rbfCount, outputCount];
for (i = 0; i < rbfCount; i++)
{
for (int j = 0; j < outputCount; j++)
{
outputLayer_weights[i, j] = Serializer.DeserializeFromString(_outputLayerWeights[i].OutputLayerWeight)[j];
}
}
classesNames = _classesNames;
iterationCount = 0;
}
private void TeachRBFNeurons(Dictionary <char, List <double[]> > classes)
{
List <RBF_teachInfo> RBFs = new List <RBF_teachInfo>();
foreach (List <double[]> matchClass in classes.Values)
{
foreach (double[] match in matchClass)
{
RBF_teachInfo rbf = new RBF_teachInfo();
rbf.expectations = match;
RBFs.Add(rbf);
}
}
rbfCount = RBFs.Count;
rbf_teached = RBFs.ToArray();
if (rbfCount == 1)
{
double sko = 0;
for (int k = 0; k < inputCount; k++)
{
sko += Math.Pow((rbf_teached[0].expectations[k]), 2);
}
rbf_teached[0].standardDeviation = Math.Sqrt(sko);
return;
}
List <double>[] deviations = new List <double> [rbfCount];
for (int i = 0; i < deviations.Length; i++)
{
deviations[i] = new List <double>();
}
for (int i = 0; i < rbfCount; i++)
{
for (int j = i + 1; j < rbfCount; j++)
{
double sko = 0;
for (int k = 0; k < inputCount; k++)
{
sko += Math.Pow((rbf_teached[i].expectations[k] - rbf_teached[j].expectations[k]), 2);
}
sko = Math.Sqrt(sko);
deviations[i].Add(sko);
deviations[j].Add(sko);
}
}
for (int i = 0; i < rbfCount; i++)
{
rbf_teached[i].standardDeviation = deviations[i].Min();
}
}
