//[Test]
public void TestCP_parz()
{
double[] w1 = {0.0, 0.0, 0.0};
double[] w2 = {0, 0, 1.0};
double[] w3 = {0, 1.0, 0};
double[] w4 = {0.0, 1.0, 1.0};
double[] w5 = {1.0, 0.0, 0};
double[] w6 = {1.0, 0, 1.0};
double[] w7 = {1.0, 1.0, 0};
double[] w8 = {1.0, 1.0, 1.0};
double[][] input = new double[][]{w1,w2,w3,w4,w5,w6,w7,w8};
/*
this._networkKohonen2.Layers[0].Alpha = 0.12;
this._networkKohonen2.Layers[0].Dim = 0;
this._networkKohonen2.Layers[0].Learn(5000,input);
*/
_kohonenLearning_parz = new KohonenLearning(this._networkKohonen_parz.Layers[0]);
_kohonenLearning_parz.LearningRate = 0.12;
_kohonenLearning_parz.LearningRadius = 0;
for(int i=0;i<5000;i++){
_kohonenLearning_parz.LearningRate *= 0.999;
_kohonenLearning_parz.RunEpoch(input);
}
WidrowHoffLearning _widrowHoff = new WidrowHoffLearning(this._networkKohonen_parz.Layers[1]);
double[] out1 = {0.0,1.0};
double[] out2 = {1.0, 0.0};
double[] out3 = {1.0, 0.0};
double[] out4 = {0.0,1.0};
double[] out5 = {1.0, 0.0};
double[] out6 = {0.0,1.0};
double[] out7 = {0.0,1.0};
double[] out8 = {1.0, 0.0};
double[][] output = new double[][]{out1, out2, out3,out4,out5, out6, out7, out8};
_widrowHoff.LearningRate = 0.6;
for(int i=0;i<200;i++){
_widrowHoff.LearningRate *= 0.99;
int index = 0;
foreach(double[] wektor in input){
this._networkKohonen_parz.Layers[0].Compute(new List<double>(wektor));
IList<double> widrowInput = this._networkKohonen_parz.Layers[0].Outputs;
double[] widrowInputArray = new double[8];
widrowInput.CopyTo(widrowInputArray,0);
// foreach(double d in widrowInputArray)
// Console.WriteLine(d + " " );
_widrowHoff.Run(widrowInputArray, output[index]);
//Console.WriteLine( _widrowHoff.Run(widrowInputArray, output[index%3]));
index++;
}
}
Console.WriteLine("wagi:");
INeuronsCollection neurony = this._networkKohonen_parz.Layers[0].Neurons;
int counter = 0;
foreach (INeuron n in neurony)
{
foreach (double val in n.Weights)
{
Console.Write(val.ToString("##.###") + " || " );
counter++;
if(counter % 3 == 0)
Console.WriteLine("");
}
Console.WriteLine("--------");
}
Console.WriteLine("----------------------WARSTWA DRUGA WAGI -----------------------------");
neurony = this._networkKohonen_parz.Layers[1].Neurons;
counter = 0;
foreach (INeuron n in neurony)
{
Console.WriteLine("THRESHOLD: " + n.Threshold);
foreach (double val in n.Weights)
{
Console.Write(val.ToString("##.###") + " || " );
counter++;
if(counter % 3 == 0)
Console.WriteLine("");
}
Console.WriteLine("--------");
}
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("000.txt"));
Console.WriteLine("000:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("001.txt"));
Console.WriteLine("001:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("010.txt"));
Console.WriteLine("010:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("011.txt"));
Console.WriteLine("011:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("100.txt"));
Console.WriteLine("100:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("101.txt"));
Console.WriteLine("101:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("110.txt"));
Console.WriteLine("110:") ;
ShowOut(this._networkKohonen_parz.Outputs);
this._networkKohonen_parz.Compute(DataLoader.ReadInputs("111.txt"));
Console.WriteLine("111:") ;
ShowOut(this._networkKohonen_parz.Outputs);
}
public void TestKohonen1()
{
//List<List<double>> input = new List<List<double>>();
/*
//wektor 1
List<double> w1 = new List<double> {1, 1, 0, 0};
//wektor 2
List<double> w2 = new List<double> {0, 0, 0, 1};
//wektor 3
List<double> w3 = new List<double> {1, 0, 0, 0};
//wekror 4
List<double> w4 = new List<double> {0, 0, 1, 1};
input.Add(w1);
input.Add(w2);
input.Add(w3);
input.Add(w4);
*/
double[] w1 = {1, 1, 0, 0};
double[] w2 = {0, 0, 0, 1};
double[] w3 = {1, 0, 0, 0};
double[] w4 = {0, 0, 1, 1};
double[][] input = new double[][]{w1,w2,w3,w4};
/*
this._networkKohonen.Layers[0].Alpha = 0.6;
this._networkKohonen.Layers[0].Learn(100,input);
*/
_kohonenLearning = new KohonenLearning(this._networkKohonen.Layers[0]);
_kohonenLearning.LearningRate = 0.13;
_kohonenLearning.LearningRadius = 0;
for(int i=0;i<4000;i++){
_kohonenLearning.RunEpoch(input);
}
this._networkKohonen.Compute(DataLoader.ReadInputs("Inputs_Kohonen.txt"));
Console.WriteLine("wagi:");
INeuronsCollection neurony = this._networkKohonen.Layers[0].Neurons;
foreach (INeuron n in neurony){
foreach (double val in n.Weights)
Console.WriteLine(val);
Console.WriteLine("--------");
}
Console.WriteLine(this._networkKohonen.Outputs[0] + " " + this._networkKohonen.Outputs[1]);
Assert.Greater(this._networkKohonen.Outputs[1],this._networkKohonen.Outputs[0] );
}
//[Test]
public void TestCP_2()
{
double[] w1 = {1.0, 0.0, 0.0, 0.0, 1.0, 0, 0, 0, 1.0};
double[] w2 = {1.0, 0, 0, 1.0, 0.0, 0.0, 0, 1.0, 1.0};
double[] w3 = {1.0, 1.0, 0, 0, 0, 1.0, 0.0, 0.0, 1.0};
double[] w4 = {1.0, 0, 1.0, 0.0, 1.0, 0, 1.0, 0, 1.0};
double[] w5 = {0, 1.0, 1.0, 0, 1.0, 0, 1.0, 1.0, 0};
double[] w6 = {1.0, 0, 0.0, 1.0, 1.0, 1.0, 0, 0, 1.0};
double[] w7 = {1, 1.0, 1.0, 1.0, 0, 0, 1.0, 0, 0};
double[] w8 = {0, 1.0, 0.0, 0, 1.0, 0, 1.0, 1.0, 1.0};
double[] w9 = {1.0, 1.0, 1.0, 0, 0, 1.0, 0, 0, 1};
double[][] input = new double[][]{w1,w2,w3,w4,w5,w6,w7,w8,w9};
/*
this._networkKohonen2.Layers[0].Alpha = 0.12;
this._networkKohonen2.Layers[0].Dim = 0;
this._networkKohonen2.Layers[0].Learn(5000,input);
*/
_kohonenLearning = new KohonenLearning(this._networkKohonen3.Layers[0]);
_kohonenLearning.LearningRate = 0.02;
_kohonenLearning.LearningRadius = 0;
for(int i=0;i<5000;i++){
_kohonenLearning.LearningRate *= 0.999;
_kohonenLearning.RunEpoch(input);
}
WidrowHoffLearning _widrowHoff = new WidrowHoffLearning(this._networkKohonen3.Layers[1]);
double[] out1 = {1.0, 0.0, 0.0};
double[] out2 = {0.0, 1.0, 0.0};
double[] out3 = {0.0, 0.0, 1.0};
double[][] output = new double[][]{out1, out2, out3};
_widrowHoff.LearningRate = 0.06;
for(int i=0;i<30;i++){
_widrowHoff.LearningRate *= 0.99;
int index = 0;
foreach(double[] wektor in input){
this._networkKohonen3.Layers[0].Compute(new List<double>(wektor));
IList<double> widrowInput = this._networkKohonen3.Layers[0].Outputs;
double[] widrowInputArray = new double[9];
widrowInput.CopyTo(widrowInputArray,0);
// foreach(double d in widrowInputArray)
// Console.WriteLine(d + " " );
_widrowHoff.Run(widrowInputArray, output[index%3]);
//Console.WriteLine( _widrowHoff.Run(widrowInputArray, output[index%3]));
index++;
}
}
Console.WriteLine("wagi:");
INeuronsCollection neurony = this._networkKohonen3.Layers[0].Neurons;
int counter = 0;
foreach (INeuron n in neurony)
{
foreach (double val in n.Weights)
{
Console.Write(val.ToString("##.###") + " || " );
counter++;
if(counter % 3 == 0)
Console.WriteLine("");
}
Console.WriteLine("--------");
}
Console.WriteLine("----------------------WARSTWA DRUGA WAGI -----------------------------");
neurony = this._networkKohonen3.Layers[1].Neurons;
counter = 0;
foreach (INeuron n in neurony)
{
foreach (double val in n.Weights)
{
Console.Write(val.ToString("##.###") + " || " );
counter++;
if(counter % 3 == 0)
Console.WriteLine("");
}
Console.WriteLine("--------");
}
this._networkKohonen3.Compute(DataLoader.ReadInputs("test_2_1.txt"));
Console.WriteLine("test 2_1:");
ShowOut(this._networkKohonen3.Outputs);
this._networkKohonen3.Compute(DataLoader.ReadInputs("test_2_2.txt"));
Console.WriteLine("test 2_2:");
ShowOut(this._networkKohonen3.Outputs);
this._networkKohonen3.Compute(DataLoader.ReadInputs("test_2_3.txt"));
Console.WriteLine("test 2_3:");
ShowOut(this._networkKohonen3.Outputs);
}