public object[,] getCluster(int contextCode, double[] I)
{
LayerF2 F2 = (LayerF2)contextField[contextCode];
object[,] cluster = new object[1, 2];
double[] T = new double[F2.Count];
double[] i = new double[I.Length * 2];
if (complementCoding)
{
i = new double[I.Length * 2];
i = complement(I);
}
else
{
i = new double[I.Length];
i = I;
}
T = F1.processInput(i, alpha, F2);
int repeatCount = 1;
Repeater: // Control shall come up here in case there are more than one members in T,
object[,] j_AND_Z_td_J = F2.processInput(T);
int j = (int)j_AND_Z_td_J[0, 0];
double[] Z_td_J = (double[])j_AND_Z_td_J[0, 1];
//Adaptation Phase
if (vigilancePass(i, Z_td_J))
{
cluster[0, 0] = Z_td_J;
cluster[0, 1] = j;
}
else
{
if (repeatCount <= T.Length)
{
T[j] = -1;
repeatCount++;
goto Repeater;
}
// MinDist F2Neuron does not resonate with this i, make a new category for it
else
{
double[] d = { -1.0 };
cluster[0, 0] = d;
cluster[0, 1] = -1;
}
}
return(cluster);
}
public object[,] feedInput(int contextCode, double[] I, int catCode)
{
object[,] output = new object[1, 2];
double[] z_td_J_new = new double[i.Length];
if (complementCoding)
{
i = complement(I);
}
else
{
i = I;
}
if (!contextField.ContainsKey(contextCode))
{
//Add Context Code
contextField.Add(contextCode, new LayerF2());
}
//Load Context
LayerF2 F2Neurons = (LayerF2)contextField[contextCode];
if (F2Neurons.Count == 0)
{
output[0, 1] = F2Neurons.AddF2Neuron(F1, i);
output[0, 0] = ((F2Neuron)F2Neurons[0]).getProtoTypeCluster();
}
else
{
double[] T = new double[F2Neurons.Count];
// Recognition Phase (Orientation Subsystem)
T = F1.processInput(i, alpha, F2Neurons);
int repeatCount = 1;
Repeater: // Control shall come up here in case there are more than one members in T,
object[,] j_AND_Z_td_J = F2Neurons.processInput(T);
int J = (int)j_AND_Z_td_J[0, 0];
double[] Z_td_J = (double[])j_AND_Z_td_J[0, 1];
//Adaptation Phase
if (vigilancePass(i, Z_td_J))
{
if (((F2Neuron)F2Neurons[J]).getMapFieldConnection().getCategory().getCode() == catCode)
{
z_td_J_new = F2Neurons.updateWeights(Z_td_J, i, J, beta);
output[0, 0] = z_td_J_new;
output[0, 1] = J;
}
else // invoke match tracking
{
double normIandZ = 0.0;
double normI = 0.0;
for (int index = 0; index < i.Length; index++)
{
normI += Math.Abs(i[index]);
normIandZ += Math.Abs((Math.Min(i[index], Z_td_J[index])));
}
double ro = Math.Round((normIandZ / normI), 4) + this.epsilon;
reSetRho(ro);
T[J] = -1;
repeatCount++;
goto Repeater;
}
}
else
{
if (repeatCount < T.Length)
{
T[J] = -1;
repeatCount++;
goto Repeater;
}
// MinDist F2Neuron does not resonate with this i, make a new category for it
else
{
if (CategoryCountLimit != -1 & F2Neurons.Count == CategoryCountLimit)
{
output[0, 1] = "-1";
output[0, 0] = i;
}
else
{
output[0, 1] = F2Neurons.AddF2Neuron(F1, i);
output[0, 0] = ((F2Neuron)F2Neurons[F2Neurons.Count - 1]).getProtoTypeCluster();
}
}
}
}
return(output);
}
