private void buttonRun_Click(object sender, EventArgs e)
{
Cursor.Current = Cursors.WaitCursor;
textBoxResult.Clear();
Matrix <double> testSetInput, testSetOutput;
CreateTestingSet(out testSetInput, out testSetOutput);
testSetPredictedOutcome = network.Forward(testSetInput);
NetworkGraphics.ToTreeView(treeView1, network);
NetworkGraphics.ToPictureBox(pictureBox1, network, 400, 0);
textBoxResult.AppendText(dataSet.ProcessTestSetResults(testSetInput, testSetOutput, testSetPredictedOutcome));
Cursor.Current = Cursors.Default;
}
private void buttonRunCase_Click(object sender, EventArgs e)
{
if (network.IsTrained && dataSet.GetType() == typeof(IrisFlowerDataSets))
{
Cursor.Current = Cursors.WaitCursor;
var caseSet = new DataSet[1];
caseSet[0] = new IrisFlower(Convert.ToDouble(textBoxSL.Text), Convert.ToDouble(textBoxSW.Text), Convert.ToDouble(textBoxPL.Text), Convert.ToDouble(textBoxPW.Text), IrisFlower.IrisSpecies.Unknown);
textBoxResult.Clear();
Matrix <double> runCaseInput = dataSet.CreatePropertiesMatrixFromSet(caseSet);
Matrix <double> predicted = network.Forward(runCaseInput);
textBoxResult.AppendText(DataSet.ClassifyOutput(predicted.Column(0), NetworkGraphics.OutputNames, caseSet[0]));
NetworkGraphics.ToTreeView(treeView1, network);
NetworkGraphics.ToPictureBox(pictureBox1, network, 400, 0);
Cursor.Current = Cursors.Default;
}
}
private void buttonExport_Click(object sender, EventArgs e)
{
String path = Path.GetDirectoryName(System.Reflection.Assembly.GetExecutingAssembly().Location);
string fileDetails = path + "\\NeuralNetworkConfig" + DateTime.Now.ToString("yyyy-dd-M--HH-mm-ss") + ".txt";
string xmlFile = path + "\\NeuralNetworkConfig" + DateTime.Now.ToString("yyyy-dd-M--HH-mm-ss") + ".xml";
string sets = path + "\\DataSet" + DateTime.Now.ToString("yyyy-dd-M--HH-mm-ss") + ".csv";
string testResult = path + "\\TestRun" + DateTime.Now.ToString("yyyy-dd-M--HH-mm-ss") + ".txt";
File.AppendAllText(fileDetails, "NEURAL NETWORK CONFIG");
File.AppendAllText(fileDetails, "\r\nLearning Rate: " + network.Alpha);
File.AppendAllText(fileDetails, "\r\nMomentum: " + network.Momentum);
File.AppendAllText(fileDetails, "\r\nEpochs used on the training: " + network.EpochsIterations);
File.AppendAllText(fileDetails, "\r\n\r\nLAYERS STRUCTURE");
foreach (var layer in network.Layers)
{
File.AppendAllText(fileDetails, "\r\n\r\nLayer " + layer.Name + " has " + layer.NeuronCount + " neurons");
File.AppendAllText(fileDetails, "\r\nActivation Function: " + layer.activationFunction.Method.Name);
if (network.UseNeuronBias)
{
for (int i = 0; i < layer.NeuronCount; i++)
{
File.AppendAllText(fileDetails, "\r\nNeuron " + i + " - Bias: " + layer.Bias[1, i]);
}
}
if (layer.Synapse != null)
{
File.AppendAllText(fileDetails, "\r\nWeights Matrix - " + layer.Synapse);
}
}
ExportDataSets(sets);
NetworkGraphics.ExportTreeViewToXml(treeView1, xmlFile);
MessageBox.Show("Files " + fileDetails + " & " + xmlFile + " SUCCESSFULY CREATED");
}
private void buttonTrain_Click(object sender, EventArgs e)
{
Cursor.Current = Cursors.WaitCursor;
Matrix <double> trainingSetInput, trainingSetOutput;
CreateTrainingSet(out trainingSetInput, out trainingSetOutput);
textBoxResult.Clear();
//////////////////////Configure Network///////////////////////////
List <int> hiddenLayersWidth = new List <int>();
hiddenLayersWidth.Add(3);
network.EpochsIterations = 1000000;
network.Alpha = 0.005;
//network.Momentum = 0.1;
network.UseNeuronBias = false;
//////////////////////////////////////////////////////////////////
var watch = System.Diagnostics.Stopwatch.StartNew();
network.TrainNetwork(trainingSetInput, trainingSetOutput, hiddenLayersWidth);
watch.Stop();
var elapsedMs = watch.ElapsedMilliseconds;
//Console.WriteLine("Elapsed Training Time (ms): " + elapsedMs); //For Debug purposes
NetworkGraphics.ToTreeView(treeView1, network);
NetworkGraphics.ToPictureBox(pictureBox1, network, 400, 0);
buttonRun.Enabled = network.IsTrained;
buttonRunCase.Enabled = network.IsTrained && (dataSet.GetType() == typeof(IrisFlowerDataSets));
buttonGenerate.Enabled = network.IsTrained && (dataSet.GetType() == typeof(IrisFlowerDataSets));
buttonExport.Enabled = network.IsTrained;
Cursor.Current = Cursors.Default;
}