public static void ToNN(this TreeView t, NeuralNetworkImplementation nn, Dictionary <int, Func <double, double> > funcDictionary)
{
t.Nodes.Clear();
TreeNode root = new TreeNode("NeuralNetwork");
int layerNum = 1;
nn.Layers.ForEach((layer) =>
{
TreeNode lnode = new TreeNode($"Layer[{layerNum}]");
lnode.Nodes.Add($"[{funcDictionary.First(f => f.Value == layer.ActivationFunction).Key.ToString()}] Activation function");
layerNum++;
int neuronNum = 1;
layer.Neurons.ForEach((neuron) =>
{
TreeNode nnode = new TreeNode($"Neuron[{neuronNum}]");
neuronNum++;
nnode.Nodes.Add("Bias: " + layer.Bias.ToString());
neuron.Dendrites.ForEach((dendrite) =>
{
TreeNode dnode = new TreeNode("Dendrite");
dnode.Nodes.Add("Weight: " + dendrite.Weight.ToString());
nnode.Nodes.Add(dnode);
});
lnode.Nodes.Add(nnode);
});
root.Nodes.Add(lnode);
});
t.Nodes.Add(root);
}
public NN()
{
InitializeComponent();
funcDictionary = new Dictionary <int, Func <double, double> >
{
{ 1, (x) => x },
{ 2, (x) => x * x },
{ 3, (x) => x > 0 ? 1 : 0 }
};
var linear = GetData <LinearLayerModel>(@"D:\Projects\NeuralNetwork-master\NN.Interpolation\source1.json");
var layers1 = LayerModelToBaseLayer.MapLinear(linear, funcDictionary);
_neuralNetwork = new NeuralNetworkImplementation(layers1, 1);
VisualiseNN();
}
private void NNTrain_Click(object sender, EventArgs e)
{
var linear = GetData <LinearLayerModel>(@"D:\Projects\NeuralNetwork-master\NN.Interpolation\source1.json");
var layers1 = LayerModelToBaseLayer.MapLinear(linear, funcDictionary);
_neuralNetwork = new NeuralNetworkImplementation(layers1, 1);
List <double> inputs = new List <double>();
List <double> outputs = new List <double>();
for (double i = 0; i < 1; i += 0.01)
{
inputs.Add(i);
outputs.Add(_function(i));
}
_neuralNetwork.Train(inputs, outputs);
VisualiseNN();
}
public static void ToNN(this PictureBox p, NeuralNetworkImplementation nn, int X, int Y)
{
int neuronWidth = 30;
int neuronDistance = 50;
int layerDistance = 50;
int fontSize = 8;
Bitmap b = new Bitmap(p.Width, p.Height);
Graphics g = Graphics.FromImage(b);
g.FillRectangle(Brushes.White, g.ClipBounds);
int y = Y;
for (int l = 0; l < nn.Layers.Count; l++)
{
Layer <double, double> layer = nn.Layers[l];
int x = X - (neuronDistance * (layer.Neurons.Count / 2));
for (int n = 0; n < layer.Neurons.Count; n++)
{
Neuron <double> neuron = layer.Neurons[n];
for (int d = 0; d < neuron.Dendrites.Count; d++)
{
// TO DO: optionally draw dendrites between neurons
}
;
g.FillEllipse(Brushes.WhiteSmoke, x, y, neuronWidth, neuronWidth);
g.DrawEllipse(Pens.Gray, x, y, neuronWidth, neuronWidth);
//g.DrawString(neuron.Akson().ToString("0.00"), new Font("Arial", fontSize), Brushes.Black, x + 2, y + (neuronWidth / 2) - 5);
g.DrawString("", new Font("Arial", fontSize), Brushes.Black, x + 2, y + (neuronWidth / 2) - 5);
x += neuronDistance;
}
;
y += layerDistance;
}
;
p.Image = b;
}
