public TestingMode(Perceptron _perceptron)
{
teachingPatterns = new TeachingPatterns();
perceptron = _perceptron;
perceptron.uploadWeights();
for (int j = 1; j <= 4; j++)
{
input = teachingPatterns.getTeachingPattern(j);
expected = input; // W tym przypdku wartości wyjściowe to wejściowe
perceptron.forwardPropagation(input);
perceptron.addSumSquaredError(expected);
string strInput = "";
for (int k = 0; k < perceptron.layers[0].input.Length; k++)
{
strInput += Convert.ToString(perceptron.layers[0].input[k]) + " ";
}
string strOutput = "";
for (int l = 0; l < perceptron.layers.Last().output.Length; l++)
{
strOutput += Convert.ToString(Math.Round(perceptron.layers.Last().output[l], 4)) + " ";
}
Console.WriteLine(String.Format("input: {0} output: {1}", strInput, strOutput));
}
perceptron.estimateSumSquaredError();
Console.WriteLine("błąd średniokwadratowy: {0}\n", perceptron.getTotalSumSquaredError());
}
public TeachingMode(Perceptron _perceptron, int _epoch, int _logJump, bool _isRandomOrderPatterns, double _learnFactor, double _momentum, double _errorLevel = 0.0)
{
perceptron = _perceptron;
perceptron.randomWeights();
for (int i = 0; i <= _epoch; i++)
{
teachingPatterns = new TeachingPatterns();
for (int j = 1; j <= 4; j++)
{
if (_isRandomOrderPatterns == true)
{
input = teachingPatterns.getRandomTeachingPattern();
}
else
{
input = teachingPatterns.getTeachingPattern(j);
}
expected = input; // W tym przypdku wartości wyjściowe to wejściowe
/**** BACK PROPAGATION ALGORITHM ****/
perceptron.backPropagation(_learnFactor, _momentum, input, expected);
/************************************/
perceptron.addSumSquaredError(expected);
}
perceptron.estimateSumSquaredError();
if (i % _logJump == 0)
{
string strInput = "";
for (int k = 0; k < perceptron.layers[0].input.Length; k++)
{
strInput += Convert.ToString(perceptron.layers[0].input[k]) + " ";
}
string strOutput = "";
for (int l = 0; l < perceptron.layers.Last().output.Length; l++)
{
strOutput += Convert.ToString(Math.Round(perceptron.layers.Last().output[l], 4)) + " ";
}
Console.WriteLine(String.Format("input: {0} output: {1}", strInput, strOutput));
perceptron.saveErrorToFile();
}
if (perceptron.getTotalSumSquaredError() <= _errorLevel || i == _epoch)
{
Console.WriteLine("błąd średniokwadratowy: {0}\n założony poziom błędu: {1}\n epoki: {2}",
perceptron.getTotalSumSquaredError(), _errorLevel, i);
Console.WriteLine(perceptron.ToString());
break;
}
perceptron.resetSumSquaredError();
}
}
public ClassificationTestingMode(Perceptron _perceptron)
{
cts = new ClassificationTeachingSet();
perceptron = _perceptron;
perceptron.uploadWeights();
for (int i = 1; i <= 3; i++)
{
for (int j = 0; j < 3; j++)
{
input = cts.getTestingPattern(i, j);
expected = cts.getOutput(i);
perceptron.forwardPropagation(input);
perceptron.addSumSquaredError(expected);
string strInput = "";
for (int k = 0; k < perceptron.layers[0].input.Length; k++)
{
strInput += Convert.ToString(perceptron.layers[0].input[k]) + " ";
}
string strExpected = "";
for (int k = 0; k < expected.Length; k++)
{
strExpected += Convert.ToString(expected[k]);
}
string strOutput = "";
for (int k = 0; k < perceptron.layers.Last().output.Length; k++)
{
strOutput += Convert.ToString(Math.Round(perceptron.layers.Last().output[k], 4)) + " ";
}
Console.WriteLine(String.Format("input: {0} expected {1} output: {2}", strInput, strExpected, strOutput));
}
perceptron.estimateSumSquaredError();
}
Console.WriteLine("błąd średniokwadratowy: {0}\n", perceptron.getTotalSumSquaredError());
}
