private void Recall(bool reconstruct)
{
string[] sp = textBox1.Text.Split(',');
if (sp.Length != 2)
{
label1.Text = "You need to enter <neuron>,<layer>!";
label1.Refresh();
return;
}
try
{
int neuron = int.Parse(sp[0]);
int layer = int.Parse(sp[1]);
string c = (layer == LAYERS.Length - 1) ? listBox1.Items[neuron].ToString() : "(not a category)";
double[] a = (reconstruct) ? new double[LAYERS[layer]] : new double[NUM_CATEGORIES];
a[neuron] = 1;
double[] r = (reconstruct) ? _network.Reconstruct(a, layer) : _network.GenerateInput(a);
Bitmap bm;
_atoi.Convert(r, out bm);
label1.Text = "Reconstructing " + c + ", length of reconstruction: " + r.Length;
label1.Refresh();
pictureBox1.Image = bm;
pictureBox1.Refresh();
}
catch (Exception ex)
{
label1.Text = ex.Message + "\n" + ex.StackTrace + "\n";
label1.Text += "Reconstruction input params invalid. neuron should be < size of layer.";
label1.Refresh();
}
}
public void Compute()
{
if (!CanCompute)
{
return;
}
double[] input = UserInput;
DeepBeliefNetwork network = Main.Network;
IDatabase database = Main.Database;
database.Normalize(input);
{
double[] output = network.GenerateOutput(input);
double[] reconstruction = network.Reconstruct(output);
NetworkOutput = (database.ToBitmap(reconstruction).ToBitmapImage());
}
if (Main.CanClassify)
{
double[] output = network.Compute(input);
int imax; output.Max(out imax);
Classification = imax;
}
}
/// <summary>
/// Computes the reconstruction error for
/// a given set of input values.
/// </summary>
///
/// <param name="inputs">The input values.</param>
///
/// <returns>The squared reconstruction error.</returns>
///
public double ComputeError(double[][] inputs)
{
double error = 0;
for (int i = 0; i < inputs.Length; i++)
{
double[] output = network.Compute(inputs[i]);
double[] reconstruct = network.Reconstruct(output);
for (int j = 0; j < inputs[i].Length; j++)
{
double e = reconstruct[j] - inputs[i][j];
error += e * e;
}
}
return(error);
}
/// <summary>
/// Implementation of the divergence operation.
/// </summary>
protected override void diverge()
{
//convert all the modalities in a single vector
double[] realSignal; double[] predictedSignal;
getConcatenatedModalities(out realSignal, out predictedSignal);
//Generate the prediction
double[] output = network.GenerateOutput(realSignal, 0);
double[] networkPrediction = network.Reconstruct(output);
//Distribute it over the Signal
setConcatenatedModalities(null, networkPrediction);
//Proceed to learning
if (!learningLocked && learningRate > 0)
{
//teacher.Run(realSignal);
//network.UpdateVisibleWeights();
}
}