public virtual NeuralNet initNet(int numberOfInputNeurons, int numberOfHiddenLayers, int numberOfNeuronsInHiddenLayer, int numberOfOutputNeurons)
{
inputLayer = new InputLayer();
inputLayer.NumberOfNeuronsInLayer = numberOfInputNeurons;
listOfHiddenLayer = new List <HiddenLayer>();
for (int i = 0; i < numberOfHiddenLayers; i++)
{
hiddenLayer = new HiddenLayer();
hiddenLayer.NumberOfNeuronsInLayer = numberOfNeuronsInHiddenLayer;
listOfHiddenLayer.Add(hiddenLayer);
}
outputLayer = new OutputLayer();
outputLayer.NumberOfNeuronsInLayer = numberOfOutputNeurons;
inputLayer = inputLayer.initLayer(inputLayer);
if (numberOfHiddenLayers > 0)
{
listOfHiddenLayer = hiddenLayer.initLayer(hiddenLayer, listOfHiddenLayer, inputLayer, outputLayer);
}
outputLayer = outputLayer.initLayer(outputLayer);
NeuralNet newNet = new NeuralNet();
newNet.InputLayer = inputLayer;
newNet.HiddenLayer = hiddenLayer;
newNet.ListOfHiddenLayer = listOfHiddenLayer;
newNet.NumberOfHiddenLayers = numberOfHiddenLayers;
newNet.OutputLayer = outputLayer;
return(newNet);
}
public virtual List <HiddenLayer> initLayer(HiddenLayer hiddenLayer, List <HiddenLayer> listOfHiddenLayer, InputLayer inputLayer, OutputLayer outputLayer)
{
List <double> listOfWeightIn = new List <double>();
List <double> listOfWeightOut = new List <double>();
List <Neuron> listOfNeurons = new List <Neuron>();
int numberOfHiddenLayers = listOfHiddenLayer.Count;
for (int hdn_i = 0; hdn_i < numberOfHiddenLayers; hdn_i++)
{
for (int neuron_i = 0; neuron_i < hiddenLayer.NumberOfNeuronsInLayer; neuron_i++)
{
Neuron neuron = new Neuron();
int limitIn = 0;
int limitOut = 0;
if (hdn_i == 0)
{ // first
limitIn = inputLayer.NumberOfNeuronsInLayer;
if (numberOfHiddenLayers > 1)
{
limitOut = listOfHiddenLayer[hdn_i + 1].NumberOfNeuronsInLayer;
}
else if (numberOfHiddenLayers == 1)
{
limitOut = outputLayer.NumberOfNeuronsInLayer;
}
}
else if (hdn_i == numberOfHiddenLayers - 1)
{ // last
limitIn = listOfHiddenLayer[hdn_i - 1].NumberOfNeuronsInLayer;
limitOut = outputLayer.NumberOfNeuronsInLayer;
}
else
{ // middle
limitIn = listOfHiddenLayer[hdn_i - 1].NumberOfNeuronsInLayer;
limitOut = listOfHiddenLayer[hdn_i + 1].NumberOfNeuronsInLayer;
}
limitIn = limitIn - 1; // bias no
limitOut = limitOut - 1;
if (neuron_i >= 1)
{
for (int k = 0; k <= limitIn; k++)
{
listOfWeightIn.Add(neuron.initNeuron());
//listOfWeightIn.add(neuron.initNeuron(k, neuron_i, 1));
}
}
for (int k = 0; k <= limitOut; k++)
{
listOfWeightOut.Add(neuron.initNeuron());
//listOfWeightOut.add(neuron.initNeuron(k, neuron_i, 2));
}
neuron.ListOfWeightIn = listOfWeightIn;
neuron.ListOfWeightOut = listOfWeightOut;
listOfNeurons.Add(neuron);
listOfWeightIn = new List <double>();
listOfWeightOut = new List <double>();
}
listOfHiddenLayer[hdn_i].ListOfNeurons = listOfNeurons;
listOfNeurons = new List <Neuron>();
}
return(listOfHiddenLayer);
}