public double[] ReturnedError(Neuron []Y,int index)
{
double[] error = new double[OutPutLayerSize];
for (int j = 0; j < OutPutLayerSize; j++)
{
error[j] = outPuts[index][j] - Y[j].OutPutValue;
}
return error;
}
public void GetInputpublic(byte[,] buffer, int RowSize, int ColSize, Neuron[] InputLayer, int outputlayersize = 0, int NoOfIteration = 0, double LearningRate = 0.1)
{
int Count = 0;
for (int j = 0; j < RowSize; j++)
{
for (int i = 0; i < ColSize; i++)
{
InputLayer[Count++].NetInput = Convert.ToDouble(buffer[j, i]);
}
}
}
public void GetInputpublic(byte[,] buffer, int RowSize, int ColSize, Neuron[] InputLayer, int outputlayersize = 0, int NoOfIteration = 0, double LearningRate = 0.1)
{
Neuron[] PCAoutPutLayer = new Neuron[outputlayersize];
PCA P = new PCA(InputLayer.Length, outputlayersize, LearningRate);
P.GetInput(buffer,RowSize,ColSize);
P.Run(NoOfIteration);
PCAoutPutLayer = P.Show();
for(int j=0;j<outputlayersize;j++)
{
InputLayer[j].NetInput = PCAoutPutLayer[j].OutPutValue;
}
}
public PCA(int inputLayerSize, int OutPutLayerSize, double LearningRate)
{
this.InputLayerSize = inputLayerSize;
this.OutPutLayerSize = OutPutLayerSize;
this.LearningRate = LearningRate;
InputLayer = new Neuron[inputLayerSize];
OutPutLayer = new Neuron[OutPutLayerSize];
for (int j = 0; j < inputLayerSize; j++)
{
InputLayer[j] = new Neuron(OutPutLayerSize);
}
for (int j = 0; j < OutPutLayerSize; j++)
{
OutPutLayer[j] = new Neuron();
}
}
public InputStyle IS;// define the way you get input from PCA or from Images !
public NetWork(int InputLayerSize, int OutPutLayerSize, int[] HiddenLayersSize, double LearningRate,InputStyle IS)
{
// Construction of network ! input Layers ! Hidden Layers ! output Layers
this.IS = IS;
Target = new Desired(OutPutLayerSize);
this.LearningRate = LearningRate;
Target.Creat();// Create Desired output for each class;
NoOfHiddenLayers = HiddenLayersSize.Length;
InputLayer = new Neuron[InputLayerSize];
OutputLayer = new Neuron[OutPutLayerSize];
HiddenLayers = new Neuron[NoOfHiddenLayers][];
// intilize Size of hidden layers;
for (int j = 0; j < NoOfHiddenLayers; j++)
{
HiddenLayers[j] = new Neuron[HiddenLayersSize[j]];
}
// initlize Input Layer With Objects To be Trained;
for (int j = 0; j < InputLayerSize; j++)
{
InputLayer[j] = new Neuron(HiddenLayers[0].Length);
}
// initlize outPut Layer with objects;
for (int j = 0; j < OutPutLayerSize; j++)
{
OutputLayer[j] = new Neuron();
}
// intilize Hidden layers with objects except last one because it depends on outputlayer;
for (int j = 0; j < NoOfHiddenLayers - 1; j++)
{
for (int i = 0; i < HiddenLayersSize[j]; i++)
{
HiddenLayers[j][i] = new Neuron(HiddenLayersSize[j + 1]);
}
}
// intilize last Layer in network with objects
for (int j = 0; j < HiddenLayersSize[NoOfHiddenLayers - 1]; j++)
{
HiddenLayers[NoOfHiddenLayers - 1][j] = new Neuron(OutPutLayerSize);
}
}
