/// <summary>
/// Конструктор с параметром обучения нейросети
/// </summary>
public NeuralNetwork(int trainigCoof)
{
_first = new InputLayer(6);
_second = new HiddenLayer(2);
_third = new OutputLayer(6);
_sFirst = new SynapseLayer(6, 2);
_sSecond = new SynapseLayer(2, 6);
this._trainigCoof = trainigCoof;
ValueCrit = 0;
}
/// <sumarry>
/// Конструктор для наследования нейросети
/// </sumarry>
public NeuralNetwork(NeuralNetwork baseNet, bool triggerTrain)
{
_first = new InputLayer(6);
_second = new HiddenLayer(2);
_third = new OutputLayer(6);
_sFirst = new SynapseLayer(6, 2);
_sSecond = new SynapseLayer(2, 6);
_trainigCoof = baseNet.TraingCoof;
_sFirst = new SynapseLayer(baseNet.SFirst, baseNet.TraingCoof, triggerTrain);
_sSecond = new SynapseLayer(baseNet.SSecond, baseNet.TraingCoof, triggerTrain);
ValueCrit = 0;
}
static public NeuralNetwork Mixing(NeuralNetwork web1, NeuralNetwork web2)
{
SynapseLayer first = new SynapseLayer(web1.SFirst.GetLength(0), web1.SFirst.GetLength(1));
SynapseLayer second = new SynapseLayer(web1.SSecond.GetLength(0), web1.SSecond.GetLength(1));
for (int i = 0; i < first.N; ++i)
{
for (int j = 0; j < first.M; ++j)
{
first.Matrix [i, j] = (web1.SFirst [i, j] + web2.SFirst [i, j]) / 2f;
}
}
for (int i = 0; i < second.N; ++i)
{
for (int j = 0; j < second.M; ++j)
{
second.Matrix [i, j] = (web1.SSecond [i, j] + web2.SSecond [i, j]) / 2f;
}
}
return(new NeuralNetwork(web1.TraingCoof, first.Matrix, second.Matrix));
}
