public static int SampleInput(myVector sampleInput, Network N)
{
float sum;
float Error = 0;
int foundID = -35;
int correctID = 25;
int Counter = 0;
ClearInputs(N);
//float NormalizationCoef = 1 / sampleInput.GetVector().Count;
float NormalizationCoef = 1.0f / 50.0f;
var WordsList = sampleInput.GetVector();
var InputLayer = N.getNetwork().Where(o => o.type == 0).ToList();
var HiddenLayer = N.getNetwork().Where(o => o.type == 1).ToList();
var OutputLayer = N.getNetwork().Where(o => o.type == 2).ToList();
//while (foundID != correctID && Counter < 20)
//{
// sum = 0;
// N = ClearInputs(N);
// Pierwsza iteracja
for (int i = 0; i < sampleInput.GetVector().Count; i++)
{
InputLayer[i].Input = (float)WordsList[i];
}
// Druga iteracja
CalculateInput(InputLayer);
SigmoidFunction(HiddenLayer, NormalizationCoef);
// Trzecia iteracja
CalculateInput(HiddenLayer);
SigmoidFunction(OutputLayer, NormalizationCoef);
// Liczenie wyniku
foundID = -OutputLayer.OrderBy(o => o.Input).Reverse().FirstOrDefault().ID - 1;
// correctID = -CorrectId(sampleInput, OutputLayer) - 1;
//
// // Dopasowywanie wag
// sum = SumInputs(OutputLayer, sum);
// Error = (OutputLayer[correctID].Input - OutputLayer[foundID].Input) / sum;
// if (foundID != correctID)
// {
// AdjustWeights(HiddenLayer, Error);
// AdjustWeights(InputLayer, Error);
// }
// Counter++;
//}
return(foundID);
}
public void InitializeNetwork()
{
classes = DataClass.CreateDataClasses(bow);
classes = TestFunctions.CreateFullSet(classes, bow);
myVector x = new myVector();
{
x = bow.GetVectorsList()[0];
WithoutHiddenLayerNetwork = NeuralConstruction.CreateDefaultNetwork(x.GetVector().Count, classes);
NeuralConstruction.SampleWeight(WithoutHiddenLayerNetwork, bow.GetVectorsList(), classes);
NeuralNetwork = NeuralConstruction.CreateNewDefaultNetwork(x.GetVector().Count, classes, 5);
}
}
public static int OldSampleInput(myVector sampleInput, Network N)
{
var vector = sampleInput.GetVector();
int id = 0;
// Zerujemy inputy
foreach (var neuron in N.getNetwork())
{
neuron.Input = 0;
}
// Pierwsza iteracja
for (int i = 0; i < vector.Count; i++)
{
// int index = vector[i];
var neuron = N.getNetwork().ElementAt(i);
neuron.Input = (float)vector[i];
}
// Druga iteracja
for (int i = 0; i < vector.Count; i++)
{
// int index = vector[i];
var neuron = N.getNetwork().ElementAt(i);
var ConnectionVector = neuron.GetConnectionsOld();
for (int j = vector.Count, k = 0; k < ConnectionVector.Count; j++, k++)
{
N.getNetwork().ElementAt(j).Input += neuron.Input * neuron.GetWeights()[k];
}
}
int connectionCount = N.getNetwork()[0].GetConnectionsOld().Count();
double max = 0;
for (int i = vector.Count; i < vector.Count + connectionCount; i++)
{
var neuron = N.getNetwork().ElementAt(i);
if (neuron.Input > max)
{
max = neuron.Input;
id = (-neuron.ID) - 1;
}
}
return(id);
}