/// <summary>
/// Uczenie warstwy wyjsciowej
/// </summary>
/// <param name="outputDesirableData"></param>
private void OutputLayerLearning(List <double[]> outputDesirableData)
{
greenMatrix = CreateGreenMatrix();
double[,] invertedGreenmatrix = Pseudoinverse.Solve(greenMatrix);
for (int i = 0; i < neuronOutputLayer.Count; i++)
{
double[] weight = Matrix.Multiply(invertedGreenmatrix, outputDesirableData[i]);
((LinearNeuron)neuronOutputLayer[i]).Weights = weight;
}
}
public static void SVDMemoryUse()
{
int sizeX = 4000, sizeY = 5000;
double[,] matrix = new double[sizeX, sizeY];
for (int i = 0; i < sizeX; i++)
{
for (int j = 0; j < sizeY; j++)
{
matrix[i, j] = i + j;
}
}
double[,] svdA = Pseudoinverse.Solve(matrix);
Console.WriteLine("Macierz wynikowa.");
Console.WriteLine(Matrix.ToString(svdA));
GC.SuppressFinalize(matrix);
GC.SuppressFinalize(svdA);
matrix = null;
svdA = null;
}
public void SVDtest()
{
double[,] a = StartMatrix();
double[,] svdA = Pseudoinverse.Solve(a);
Console.WriteLine("Macierz wynikowa.");
Console.WriteLine(Matrix.ToString(svdA));
svdA = Pseudoinverse.Solve(svdA);
Console.WriteLine("Macierz po kolejnej operacji SVD wynikowa.");
Console.WriteLine(Matrix.ToString(svdA));
for (int i = 0; i < svdA.GetLength(0); i++)
{
for (int j = 0; j < svdA.GetLength(1); j++)
{
if (Math.Abs(svdA[i, j] - a[i, j]) > 0.05)
{
Assert.IsTrue(false);
}
}
}
Assert.IsTrue(true);
}