public TrainingResults Train(Matrix<double> trainingData, Matrix<double> trainingClasses, Matrix<double> validation, Matrix<double> validationClasses, int epochs = 500)
{
var ret = new TrainingResults();
for (int epoch = 0; epoch < epochs; epoch++)
{
var newWeights = CalcNewWeights(trainingData, trainingClasses);
hiddenLayer = newWeights.Item1;
outputLayer = newWeights.Item2;
if (epoch % 10 == 0)
{
double s, p;
CalcSquaredError(trainingData, trainingClasses, out s, out p);
ret.TrainingSquaredError.Add(s);
ret.TrainingError.Add(p);
if (validation != null && validationClasses != null)
{
CalcSquaredError(validation, validationClasses, out s, out p);
ret.ValidationSquaredError.Add(s);
ret.ValidationError.Add(p);
}
if (epoch % 100 == 0)
Console.WriteLine("Epoch {0}, Error {1}, {2}", epoch, s, p * 100);
}
}
return ret;
}
public void Train()
{
new TestScheduler().With(
async scheduler =>
{
viewModel = new TrainingViewModel(dataHandler.Object, dataViewModel.Object, trainingManager.Object, manager);
var collection = new ObservableCollection <TrainedTreeData>();
var tree = TrainedTreeData.ConstructFromDocuments(new DocumentSet {
Document = new[] { new DocumentDefinition {
Labels = new[] { "test" }
} }
});
collection.Add(tree);
dataViewModel.Setup(item => item.Result).Returns(documentSet);
dataViewModel.Setup(item => item.SelectedItems).Returns(collection);
trainingManager.Setup(item => item.Train(It.IsAny <DocumentSet>(), It.IsAny <TrainingHeader>(), It.IsAny <CancellationToken>()))
.Returns(Task.FromResult(training));
TrainingResults result = null;
viewModel.Perform.Subscribe(
results =>
{
result = results;
});
scheduler.AdvanceByMs(500);
await viewModel.Perform.Execute();
scheduler.AdvanceByMs(500);
Assert.AreEqual(training, result);
});
}
protected void ProcessingTrainingResult(TrainingResults result)
{
switch (result)
{
case TrainingResults.CONVERGENCE_NOT_REACHED:
{
MessageBox.Show(this, "Convergence to precision target not reached.", "Training result",
MessageBoxButtons.OK, MessageBoxIcon.Warning);
break;
}
case TrainingResults.INTERNAL_ERROR:
{
MessageBox.Show(this, "Internal error occurred.", "Training result", MessageBoxButtons.OK,
MessageBoxIcon.Error);
break;
}
case TrainingResults.MODEL_INVALID:
{
MessageBox.Show(this, "The data set is not valid.", "Training result", MessageBoxButtons.OK,
MessageBoxIcon.Error);
break;
}
case TrainingResults.NOT_VALIDATED:
{
MessageBox.Show(this,
string.Format(
"Convergence has been reached. Network trained! But not validated, no validation data set present in the model.{0}Trained in {1} - Mse: {2}",
Environment.NewLine, core.TrainingEpoch(), core.LastMse().ToString("F5")), "Training result",
MessageBoxButtons.OK, MessageBoxIcon.Warning);
break;
}
case TrainingResults.TRAINING_SUCCESS:
{
MessageBox.Show(this,
string.Format(
"Convergence has been reached. Network trained and validated!{0}Trained in {1} - Mse: {2}",
Environment.NewLine, core.TrainingEpoch(), core.LastMse().ToString("F5"), "Training result",
MessageBoxButtons.OK, MessageBoxIcon.Exclamation));
break;
}
case TrainingResults.VALIDATION_FAIL:
{
MessageBox.Show(this,
string.Format(
"Convergence has been reached but the network hasn't passed the validation test.{0}Trained in {1} - Mse: {2}",
Environment.NewLine, core.TrainingEpoch(), core.LastMse().ToString("F5")), "Training result",
MessageBoxButtons.OK, MessageBoxIcon.Warning);
break;
}
}
}
public void Setup()
{
dataViewModel = new Mock <IDataSelectViewModel>();
trainingViewModel = new Mock <ITrainingViewModel>();
seletableViewModel = new Mock <ISelectableViewModel>();
dataViewModel.Setup(item => item.Select).Returns(seletableViewModel.Object);
fileMonitorFactory = new Mock <IFileMonitorFactory>();
fileFolder = new Mock <IFolderBrowserDialogService>();
training = TestConstants.GetTrainingResults();
}
public IFileMonitor Create(string path, TrainingResults training)
{
Guard.NotNullOrEmpty(() => path, path);
Guard.NotNull(() => training, training);
logger.Debug("Create <{0}>", path);
return(new FileMonitor(
new FileWatcher(path, documentParser.Supported),
new LearnedClassifier(documentParser, new SvmTestClient(training.DataSet, training.Model)),
new PdfPreviewCreator()));
}
protected void DoEpoch(IEnumerable <DataSetItem> trainItems)
{
StartEpoch();
foreach (DataSetItem item in trainItems)
{
TrainingResults result = Network.Train(item);
ItemTrained(result);
UpdateSnapshot();
}
}
public void Setup()
{
manager = new StateManager();
training = TestConstants.GetTrainingResults();
dataViewModel = new Mock <IDataSelectViewModel>();
trainingManager = new Mock <ITrainingManager>();
dataHandler = new Mock <IDataHandler <TrainingResults> >();
documentSet = new DocumentSet();
documentSet.Document = new[] { new DocumentDefinition {
Labels = new[] { "Test" }
} };
}
private void ItemTrained(TrainingResults result)
{
if (result.Correct)
{
epochCorrect++;
}
epochTotal++;
if (epochTotal % 10 == 0)
{
UpdateStatus();
}
}
public void Construct()
{
var arff = ArffDataSet.CreateSimple("Test");
var header = TrainingHeader.CreateDefault();
var model = new Model();
Assert.Throws <ArgumentNullException>(() => new TrainingResults(null, header, arff));
Assert.Throws <ArgumentNullException>(() => new TrainingResults(model, null, arff));
Assert.Throws <ArgumentNullException>(() => new TrainingResults(model, header, null));
var instance = new TrainingResults(model, header, arff);
Assert.IsNotNull(instance.Header);
Assert.IsNotNull(instance.Model);
}
void Awake()
{
// PlayerPrefs.DeleteAll();
mInstance = this;
DontDestroyOnLoad(this);
// serverManager = GetComponent<ServerManager>();
userData = GetComponent <UserData>();
trainingData = GetComponent <TrainingData>();
capitulosData = GetComponent <CapitulosData>();
trainingResults = GetComponent <TrainingResults>();
dateData = GetComponent <DateData>();
resultsData = GetComponent <ResultsData>();
usersData = GetComponent <UsersData>();
firebaseAuthManager = GetComponent <FirebaseAuthManager>();
}
public void Setup()
{
var dataSet = ArffDataSet.CreateSimple("Test");
var model = new Model();
model.NumberOfClasses = 2;
model.ClassLabels = null;
model.NumberOfSVPerClass = null;
model.PairwiseProbabilityA = null;
model.PairwiseProbabilityB = null;
model.SupportVectorCoefficients = new double[1][];
model.Rho = new double[1];
model.Rho[0] = 0;
model.Parameter = new Parameter();
instance = new TrainingResults(model, TrainingHeader.CreateDefault(), dataSet);
}
protected void toolStripMenuItemClickHandler(object sender, EventArgs e)
{
try
{
if (string.Equals((sender as ToolStripMenuItem).Name, "generateExampleLayoutToolStripMenuItem"))
{
var dialog = new SaveFileDialog();
dialog.Filter = "csv|*.csv";
dialog.Title = "Generate a set layout";
dialog.ShowDialog(this);
if (string.IsNullOrEmpty(dialog.FileName))
{
return;
}
var text = Export.GenerateLayout();
using (var w = new StreamWriter(dialog.FileName))
{
w.Write(text);
}
}
if (string.Equals((sender as ToolStripMenuItem).Name, "cSVToolStripMenuItem"))
{
// Load data set from a csv file
OpenFileDialog dialog = new OpenFileDialog();
dialog.Filter = "csv|*.csv";
dialog.Title = "Set file";
dialog.ShowDialog();
string path = dialog.FileName;
if (string.IsNullOrWhiteSpace(path))
{
return;
}
GuiBehavior(AppStatus.BUSY);
int linesCount = SetLoader.LoadFromCsv(path);
GuiBehavior(AppStatus.READY);
logger.InfoFormat("Set loaded: {0}, read {1} lines", dialog.FileName, linesCount);
}
if (string.Equals((sender as ToolStripMenuItem).Name, "setToolStripMenuItem"))
{
// Open set gui to view/edit the patterns
SetUi setUi = new SetUi();
setUi.ShowDialog(this);
}
if (string.Equals((sender as ToolStripMenuItem).Name, "configurationToolStripMenuItem"))
{
// Open the options gui
ModelOptionsUi configGUI = new ModelOptionsUi();
configGUI.ShowDialog(this);
// Rebuild network
AppCoreSetup();
}
if (string.Equals((sender as ToolStripMenuItem).Name, "startWorkBenchToolStripMenuItem"))
{
GuiBehavior(AppStatus.BUSY);
TrainingResults result = TrainingResults.INTERNAL_ERROR;
ThreadStart worker = delegate
{
// Run network training
result = core.Train(NinjectBinding.GetKernel.Get <SetModel>()).Result;
};
Thread waitTraining = new Thread(worker);
waitTraining.Start();
while (waitTraining.IsAlive)
{
Application.DoEvents();
}
ProcessingTrainingResult(result);
GuiBehavior(AppStatus.READY);
}
if (string.Equals((sender as ToolStripMenuItem).Name, "exitToolStripMenuItem"))
{
// Dispose modules
// Before exit release the application resources
this.Dispose();
Application.Exit();
}
if (string.Equals((sender as ToolStripMenuItem).Name, "exportToolStripMenuItem"))
{
var dialog = new SaveFileDialog();
dialog.Filter = "csv|*.csv";
dialog.Title = "Export serie to file";
dialog.ShowDialog(this);
if (string.IsNullOrEmpty(dialog.FileName))
{
return;
}
using (StreamWriter w = new StreamWriter(dialog.FileName))
{
switch (this.tabCtrl.SelectedIndex)
{
case 0:
{
Export.SeriesToFile(w, core.CurrentMse.ToArray());
break;
}
case 1:
{
Export.SeriesToFile(w, core.CurrentWeights.ToArray());
break;
}
case 2:
{
Export.TargetOutputsToFile(w, core.CurrentOutputs);
break;
}
}
}
}
}
catch (Exception exception)
{
ExceptionManager.LogAndShowException(exception, "Error", logger);
}
}
private void ItemTrained(TrainingResults result)
{
if (result.Correct)
{
epochCorrect++;
}
epochTotal++;
if (epochTotal % 10 == 0)
{
UpdateStatus();
}
}
