/// <summary>
/// Классификатор
/// </summary>
/// <param name="inpDim">Размерность входа</param>
public Classifier(int inpDim)
{
InpDim = inpDim;
net.Add(new FullBipolyareSigmoid(InpDim, 5));
net.Add(new Softmax(2));
mNNW = new MenegerNNW(net, vectorClasses);
}
/// <summary>
/// Данные для визуализации
/// </summary>
/// <param name="n">Число классов</param>
/// <returns></returns>
public Vector[] DataVisual(int n)
{
Net net = new Net(rnd);
net.Add(new LinearLayer(this[0].InpVector.N, 2));
net.Add(new Softmax(n));
net.LerningRate = 0.0001;
net.Moment = 0;
MenegerNNW mnnw = new MenegerNNW(net, this);
mnnw.Train(2);
Vector[] vects = new Vector[2 * n];
Vector nnwOut;
List <double>[] x = new List <double> [n];
List <double>[] y = new List <double> [n];
for (int i = 0; i < n; i++)
{
x[i] = new List <double>();
y[i] = new List <double>();
}
for (int i = 0; i < Count; i++)
{
for (int j = 0; j < n; j++)
{
nnwOut = net._layers[0].Output(this[i].InpVector);
if (this[i].ClassMark == j)
{
x[j].Add(nnwOut[0]);
y[j].Add(nnwOut[1]);
}
}
}
for (int i = 0, k = 0; i < n; i++)
{
vects[k++] = Vector.ListToVector(x[i]);
vects[k++] = Vector.ListToVector(y[i]);
}
return(vects);
}
