/// <summary>
/// Create new struct.
/// </summary>
/// <param name="numNeurons">Number of neurons in layer.</param>
/// <param name="recurrin">Are neurons in layer recurring.</param>
/// <param name="activeFunc">Neuron activation function.</param>
public NeuralNetworkLayer(int numNeurons, bool recurrin, NeuralNetwork.NeuronActivationFunction activeFunc)
{
numberOfNeurons = numNeurons;
recurring = recurrin;
activationFunction = activeFunc;
biases = null;
}
/// <summary>
/// Create new struct from existing.
/// </summary>
/// <param name="src">Existing layer to copy.</param>
public NeuralNetworkLayer(NeuralNetworkLayer src)
{
numberOfNeurons = src.numberOfNeurons;
recurring = src.recurring;
activationFunction = src.activationFunction;
biases = null;
}
//sample 1d function points and return array of points
/// <summary>
/// Sample 1D function from xMin to xMax building array of 2D points.
/// </summary>
/// <param name="numSamples"></param>
/// <param name="func"></param>
/// <param name="xMin"></param>
/// <param name="xMax"></param>
/// <returns>Array of 2D points.</returns>
public static float[][] SampleFunction(int numSamples, NeuralNetwork.NeuronActivationFunction func, float xMin, float xMax)
{
float[][] res = new float[numSamples][];
for (int i = 0; i < numSamples; i++)
{
float sx = (i / (float)(numSamples - 1)) * (xMax - xMin) + xMin;
res[i] = new float[] {
sx,
func(sx)
};
}
return(res);
}
public static int GetActivationFunctionID(NeuralNetwork.NeuronActivationFunction activeFunc)
{
if (activeFunc == Identity_ActivationFunction)
{
return(0);
}
if (activeFunc == Rectifier_ActivationFunction)
{
return(1);
}
if (activeFunc == Sin_ActivationFunction)
{
return(2);
}
if (activeFunc == Cos_ActivationFunction)
{
return(3);
}
if (activeFunc == Tan_ActivationFunction)
{
return(4);
}
if (activeFunc == Tanh_ActivationFunction)
{
return(5);
}
if (activeFunc == Sinh_ActivationFunction)
{
return(6);
}
if (activeFunc == Exp_ActivationFunction)
{
return(7);
}
if (activeFunc == Sigmoid_ActivationFunction)
{
return(8);
}
if (activeFunc == Sqrt_ActivationFunction)
{
return(9);
}
if (activeFunc == Pow2_ActivationFunction)
{
return(10);
}
return(-1);
}
public static string GetActivationFunctionGLSLName(NeuralNetwork.NeuronActivationFunction activeFunc)
{
if (activeFunc == Identity_ActivationFunction)
{
return("");
}
if (activeFunc == Rectifier_ActivationFunction)
{
return("rectifier");
}
if (activeFunc == Sin_ActivationFunction)
{
return("sin");
}
if (activeFunc == Cos_ActivationFunction)
{
return("cos");
}
if (activeFunc == Tan_ActivationFunction)
{
return("tan");
}
if (activeFunc == Tanh_ActivationFunction)
{
return("tanh");
}
if (activeFunc == Sinh_ActivationFunction)
{
return("sinh");
}
if (activeFunc == Exp_ActivationFunction)
{
return("exp");
}
if (activeFunc == Sigmoid_ActivationFunction)
{
return("sigmoid");
}
if (activeFunc == Sqrt_ActivationFunction)
{
return("sqrt");
}
if (activeFunc == Pow2_ActivationFunction)
{
return("pow2");
}
return("");
}
public void LoadStructure(Stream s)
{
numberOfNeurons = Utils.IntFromStream(s);
recurring = s.ReadByte() == 1;
activationFunction = Utils.GetActivationFunctionFromID(Utils.IntFromStream(s));
}
