public void Run()
{
int counter = 0;
Parallel.ForEach(Items, file =>
{
for (int nn = 1; nn < NeuronNumber; nn++) //neuron number
{
for (int i = 0; i < RepeatForEachFile; i++) //learning tests for file with x neurons in network
{
try
{
NBN nbn = new NBN(nn, file);
var result = nbn.Run(1);
DateTime d = DateTime.Now;
History h = new History();
h.Name = String.Format("{0}_{1}_{2}_{3}_{4}_{5}_{6}_{7}", Path.GetFileNameWithoutExtension(result.Filename),
d.Year, d.Month, d.Day, d.Hour, d.Minute, d.Second, d.Millisecond);
h.Data = marekbar.Xml.Serialize <LearnResult>(result);
h.Insert();
if (PDF)
{
try
{
string filename = Common.Folder.Data + "\\" + h.Name + ".pdf";
PDFGenerate data = new PDFGenerate();
data.Filename = filename;
data.Result = result;
data.ChartFilename = GeneratePlot(result.RMSE[0], Path.GetFileNameWithoutExtension(result.Filename));
HistoryPDF pdf = new HistoryPDF(data.Result, data.ChartFilename, true);
pdf.Save(data.Filename);
}
catch
{ //don't care about it
}
}
}
catch (Exception ex)
{
ex.GetError();
}
finally
{
counter++;
if (OnUpdate != null)
{
OnUpdate(counter);
}
}
}
}
});
}
private void RunLearningProcess(bool IsResearch = false)
{
try
{
AppSetting app = new AppSetting();
if (inputDataFilename == "")
{
OpenDataFile();
}
if (inputDataFilename == "")
{
return;
}
blockInterface();
NBN nbn = new NBN(app.NeuronNumber, inputDataFilename);
nbn.NBN_Gain = app.Gain;
nbn.NBN_Activation = app.ActivationFunction;
nbn.NBN_Topography = app.TopologyType;
nbn.Threshold = app.Threshold;
nbn.IsClassification = app.IsClassification;
nbn.IsResearchMode = IsResearch;
nbn.MatLabCompareDataFolder = MatLabCompareDataFolder;
string tmpMLF = MatLabCompareDataFolder;
MatLabCompareDataFolder = "";//thanks got it
nbn.OnErrorChange += (s, error) =>
{
SetText(error.Format());
};
if (app.ShowConsole)
{
nbn.OnDebug += (s, msg) =>
{
SetDebug(msg);
};
}
if (app.DontDrawChart == false)
{
nbn.OnChartUpdate += (s, title, x, y) =>
{
SetChart(title, x, y);
};
}
if (app.ShowConsole)
{
console = new DebugConsole();
console.Show();
console.Hide();
}
chart.Series.Clear();
nbn.settings = new NeuralNetworkSettings(app.MaxIterations, app.MU, app.MUL, app.MUH, app.Scale, app.MaxError);
var result = nbn.Run(app.LearnTrials);
if (tmpMLF.Length > 0)
{
string file = tmpMLF + "\\wyniki_z_nbn_csharp.txt";
if (File.Exists(file))
{
File.Delete(file);
}
using (var w = new System.IO.StreamWriter(file, true))
{
w.WriteLine("Dane: " + Path.GetFileNameWithoutExtension(result.Filename));
w.WriteLine("Liczba neuronów: " + result.Info.nn.ToString());
w.WriteLine(string.Format("Średnie RMSE uczenia: {0}", result.AverageLearnRMSE));
w.WriteLine(string.Format("Średnie RMSE testowania: {0}", result.AverageTestRMSE));
w.WriteLine(string.Format("Średnie czas uczenia: {0}", result.AverageLearnTime));
w.WriteLine(string.Format("Średnie czas testowania: {0}", result.AverageTestTime));
w.WriteLine(string.Format("Odchylenie standardowe uczenia: {0}", double.IsNaN(result.StandardDeviation) ? 0 : result.StandardDeviation));
w.WriteLine(string.Format("Odchylenie standardowe testowania: {0}", double.IsNaN(result.StandardDeviationTest) ? 0 : result.StandardDeviationTest));
w.WriteLine(string.Format("Wskaźnik sukcesu: {0}", result.SuccessRate));
}
}
std.Text = string.Format("Odchylenie standardowe uczenia: {0} oraz testowania: {1}", double.IsNaN(result.StandardDeviation) ? 0 : result.StandardDeviation, double.IsNaN(result.StandardDeviationTest) ? 0 : result.StandardDeviationTest);
String runResult = String.Format(Resource.Inst.Get("r29"),
result.AverageLearnRMSE, result.AverageTestRMSE, result.Settings.MaxError);
runResult += " " + String.Format(Resource.Inst.Get("r167"),
(result.AverageLearnRMSE <= result.Settings.MaxError ? Resource.Inst.Get("r169") : Resource.Inst.Get("r170")));
runResult += ", " + String.Format(Resource.Inst.Get("r168"),
(result.AverageTestRMSE <= result.Settings.MaxError ? Resource.Inst.Get("r169") : Resource.Inst.Get("r170")));
try
{
times.Text = String.Format(String.Format(Resource.Inst.Get("r172"), result.SuccessRate)) + ", " + String.Format(Resource.Inst.Get("r171"),
result.AverageLearnTime.TrimEnd(new char[] { '0', '.' }),
result.AverageTestTime.TrimEnd(new char[] { '0', '.' }));
}
catch (Exception ex)
{
ex.ToLog();
}
status.Text = runResult;
if (app.AutoSaveLearningResults)
{
String file = Common.File.GetXmlFileNameForHistory();
DateTime d = DateTime.Now;
History h = new History();
h.Name = String.Format("{0}_{1}_{2}_{3}_{4}_{5}_{6}", Path.GetFileNameWithoutExtension(result.Filename),
d.Year, d.Month, d.Day, d.Hour, d.Minute, d.Second);
h.Data = marekbar.Xml.Serialize <LearnResult>(result);
h.Insert();
}
if (app.ShowConsole)
{
console.Show();
}
toolSSE.Items.Clear();
for (int i = 0; i < result.SSE.Count; i++)
{
toolSSE.Items.Add(Resource.Inst.Get("r182") + (i + 1).ToString());
int position = 1;
foreach (var sse in result.SSE[i])
{
toolSSE.Items.Add(String.Format(Resource.Inst.Get("r183"), position, Math.Round(sse, 4).ToString().Replace(",", ".")));
position++;
}
}
toolRMSE.Items.Clear();
for (int i = 0; i < result.RMSE.Count; i++)
{
toolRMSE.Items.Add(Resource.Inst.Get("r184") + (i + 1).ToString());
int position = 1;
foreach (var rmse in result.RMSE[i])
{
toolRMSE.Items.Add(String.Format(Resource.Inst.Get("r185"), position, Math.Round(rmse, 4).ToString().Replace(",", ".")));
position++;
}
}
}
catch (Exception ex)
{
info = ex.ToLog().Message;
}
finally
{
unblockInterface();
}
}
private void testForData_Click(object sender, EventArgs e)
{
if (worker != null && worker.IsBusy)
{
info = Resource.Inst.Get("r189");
return;
}
var file = Common.Dialog.Open(Resource.Inst.Get("r190"), Resource.Inst.Get("r191"), "dat");
if (file.Length > 0 && File.Exists(file))
{
RepeatWindowChoice rwc = new RepeatWindowChoice();
rwc.Text = Resource.Inst.Get("r192");
rwc.FormClosing += (owner, args) =>
{
var w = (RepeatWindowChoice)owner;
if (w.Repeat)
{
var repeats = w.NumberOfRepeats;
NeuronNumberWindow nnw = new NeuronNumberWindow();
nnw.FormClosing += (nnnwOwner, nnwArgs) => {
var nw = (NeuronNumberWindow)nnnwOwner;
if (nw.IsSelected)
{
var neurons = nw.NumberOfNeurons;
worker = new BackgroundWorker();
worker.WorkerReportsProgress = true;
worker.WorkerSupportsCancellation = true;
worker.DoWork += (a, b) =>
{
var data = ((String)b.Argument).Split(";".ToCharArray());
var inputFile = data[0];
var numberRepeats = int.Parse(data[1]);
var maxNeurons = int.Parse(data[2]);
List <LearnResult> results = new List <LearnResult>();
int counter = 0;
for (int nn = 1; nn < maxNeurons; nn++)
{
for (int r = 0; r < numberRepeats; r++)
{
NBN nbn = new NBN(nn, file);
results.Add(nbn.Run(1));
((BackgroundWorker)a).ReportProgress(++counter);
}
}
List <ReportData> items = new List <ReportData>();
for (int nn = 2; nn <= maxNeurons; nn++)
{
var list = results.Where(q => q.Info != null && q.Info.nn == nn).ToList();
var avgLearnRMSE = list.Average(q => q.AverageLearnRMSE);
var avgTestRMSE = list.Average(q => q.AverageTestRMSE);
ReportData r = new ReportData();
r.LearnData = file;
r.LearnRMSE = avgLearnRMSE.ToString();
r.NeuronNumber = nn.ToString();
r.TestRMSE = avgTestRMSE.ToString();
r.Trials = numberRepeats.ToString();
r.LearnTime = list.GetLearnTime();
r.TestTime = list.GetTestTime();
items.Add(r);
}
b.Result = items;
};
worker.ProgressChanged += (a, b) =>
{
info = b.ProgressPercentage.ToString() + " z " + maxTmp;
};
worker.RunWorkerCompleted += (a, b) =>
{
info = Resource.Inst.Get("r193");
var report = ((List <ReportData>)b.Result);
GridWindow gw = new GridWindow();
gw.SetData(Resource.Inst.Get("r194"), report);
gw.ShowDialog();
};
var arguments = String.Format("{0};{1};{2}", file, repeats, neurons);
maxTmp = (repeats * neurons).ToString();
worker.RunWorkerAsync(arguments);
}
else
{
info = Resource.Inst.Get("r195");
}
};
nnw.ShowDialog();
}
else
{
info = Resource.Inst.Get("r195");
}
};
rwc.ShowDialog();
}
else
{
info = Resource.Inst.Get("r196");
}
}
