/// <summary>
/// Train neural network
/// </summary>
/// <param name="input">Input data</param>
/// <param name="output">Output data</param>
/// <param name="rounds">Optional number of rounds (defaults to 1)</param>
/// <returns></returns>
public bool Train(NeuralData input, NeuralData output, long rounds = 1)
{
if ((input.Count != this.Layers[0].Neurons.Count) || (output.Count != this.Layers[this.Layers.Count - 1].Neurons.Count))
{
return(false);
}
for (long r = 0; r < rounds; r++)
{
Run(input);
for (int i = 0; i < this.Layers[this.Layers.Count - 1].Neurons.Count; i++)
{
Neuron neuron = this.Layers[this.Layers.Count - 1].Neurons[i];
neuron.Delta = neuron.Value * (1 - neuron.Value) * (output[i] - neuron.Value);
for (int j = this.Layers.Count - 2; j >= 1; j--)
{
for (int k = 0; k < this.Layers[j].Neurons.Count; k++)
{
Neuron n = this.Layers[j].Neurons[k];
n.Delta = n.Value *
(1 - n.Value) *
this.Layers[j + 1].Neurons[i].Dendrites[k].Weight *
this.Layers[j + 1].Neurons[i].Delta;
}
}
}
for (int i = this.Layers.Count - 1; i >= 1; i--)
{
for (int j = 0; j < this.Layers[i].Neurons.Count; j++)
{
Neuron n = this.Layers[i].Neurons[j];
n.Bias = n.Bias + (this.LearningRate * n.Delta);
for (int k = 0; k < n.Dendrites.Count; k++)
{
n.Dendrites[k].Weight = n.Dendrites[k].Weight + (this.LearningRate * this.Layers[i - 1].Neurons[k].Value * n.Delta);
}
}
}
++this.TrainingRounds;
}
return(true);
}
/// <summary>
/// Internal function to return an array of double from an input string
/// </summary>
/// <param name="input"></param>
/// <returns></returns>
private static double[] StringToNeural(string input)
{
var nd = new NeuralData();
foreach (var c in input.ToCharArray())
{
string ch = c.ToString();
double d;
double.TryParse(ch, out d);
nd.Add(d);
}
return(nd.ToArray());
}
/// <summary>
/// Executes the neural network
/// </summary>
/// <param name="input">Input data</param>
/// <returns></returns>
public NeuralData Run(NeuralData input)
{
if (input.Count != this.Layers[0].NeuronCount)
{
return(null);
}
for (int l = 0; l < Layers.Count; l++)
{
Layer layer = Layers[l];
for (int n = 0; n < layer.Neurons.Count; n++)
{
Neuron neuron = layer.Neurons[n];
if (l == 0)
{
neuron.Value = input[n];
}
else
{
neuron.Value = 0;
for (int np = 0; np < this.Layers[l - 1].Neurons.Count; np++)
{
neuron.Value = neuron.Value + this.Layers[l - 1].Neurons[np].Value * neuron.Dendrites[np].Weight;
}
neuron.Value = Activate(this.Activation, neuron.Value + neuron.Bias);
}
}
}
Layer last = this.Layers[this.Layers.Count - 1];
int numOutput = last.Neurons.Count;
double[] output = new double[numOutput];
for (int i = 0; i < last.Neurons.Count; i++)
{
output[i] = last.Neurons[i].Value;
}
return(new NeuralData(output));
}
public Form1()
{
InitializeComponent();
nn nw = new nn(1.25, new int[] { 2, 2, 1 }, Activation.SIGMOID, "TestNet");
//nw = nw.Load(@"c:\tmp\test_multip.xml");
//nw.SaveImage(@"c:\tmp\nn.jpg");
for (long i = 0; i < 500000; i++)
{
nd outx = nw.OutputDataInt(0);
nd inpx = new nd(0, 0);
nw.Train(inpx, outx, 5);
outx = nw.OutputDataInt(1);
inpx = new nd(0, 1);
nw.Train(inpx, outx, 5);
outx = nw.OutputDataInt(1);
inpx = new nd(1, 0);
nw.Train(inpx, outx, 5);
outx = nw.OutputDataInt(0);
inpx = new nd(1, 1);
nw.Train(inpx, outx, 5);
}
MessageBox.Show("Training rounds so far: " + nw.TrainingRounds.ToString());
// *****************************************************************
nd outp = nw.Run(new nd(1, 1));
MessageBox.Show(outp.BinaryToInt().ToString());
//nw.Save();
}
/// <summary>
/// Extenstion to NeuralData for quick conversion and initialization of integer output value
/// </summary>
/// <param name="nw"></param>
/// <param name="number"></param>
/// <returns></returns>
public static NeuralData OutputDataInt(this NeuralNetwork nw, long number)
{
return(new NeuralData(NeuralData.ConvertToBinary(number, nw.OutputNeuronsCount)));
}
/// <summary>
/// Extension to NeuralData for quick value conversion toward int
/// </summary>
/// <param name="nd"></param>
/// <returns></returns>
public static int BinaryToInt(this NeuralData nd)
{
return(NeuralData.BinaryToInt(nd.ToArray()));
}