/// <summary>
/// Thread for training the decisition tree
/// </summary>
private void TrainingThread()
{
System.Console.WriteLine("Inside the training!!!!!!!!!!!!");
List <FlowFeature> trainingFeatures;
// List<FlowFeature> trainingFeatures;
//MySqlDao dao = null;
try
{
//dao = new MySqlDao();
// set data table
//Flow.SetLabelTable(dao.GetFlowLabels());
string traceFile = _openTrainingSetDlg.FileName;
System.Console.WriteLine("the training file is:" + traceFile);
if (File.Exists(traceFile) == false)
{
throw new FileNotFoundException("Trace file " + traceFile + " doesn't exist.");
}
NMParser parser = new NMParser();
IEnumerable <FlowFeature> allFeatures = GetFlowFeaturesFromTraceFile(parser, traceFile, -1);
trainingFeatures = allFeatures.ToList();
List <Attribute> attributes = new List <Attribute>();
attributes.Add(new NumericalAttribute("PX"));
attributes.Add(new NumericalAttribute("APL"));
attributes.Add(new NumericalAttribute("PV"));
attributes.Add(new NumericalAttribute("DPL"));
attributes.Add(new NumericalAttribute("PPS"));
attributes.Add(new IdSymbolicAttribute("Protocol", new List <string>()
{
"TCP", "UDP", "Mixed"
}));
attributes.Add(new NumericalAttribute("FPS"));
attributes.Add(new NumericalAttribute("AB"));
attributes.Add(new NumericalAttribute("TBT"));
attributes.Add(new NumericalAttribute("BS"));
attributes.Add(new NumericalAttribute("PS"));
attributes.Add(new NumericalAttribute("NNP"));
attributes.Add(new NumericalAttribute("NSP"));
attributes.Add(new NumericalAttribute("PSP"));
attributes.Add(new NumericalAttribute("Duration"));
attributes.Add(new NumericalAttribute("AIT"));
attributes.Add(new NumericalAttribute("IOPR"));
attributes.Add(new NumericalAttribute("Reconnect"));
attributes.Add(new IdSymbolicAttribute("Type", Flow2.GetFlowTypeNames()));
// System.Diagnostics.Debug.WriteLine("TrainingThread1");
AttributeSet attrSet = new AttributeSet(attributes);
ItemSet itemSet = new ItemSet(attrSet);
Dictionary <int, int> maliciousFlowCounter = new Dictionary <int, int>();
int value;
foreach (FlowFeature feature in trainingFeatures)
{
List <AttributeValue> attrVals = new List <AttributeValue>();
attrVals.Add(new KnownNumericalValue(feature.PX));
attrVals.Add(new KnownNumericalValue(feature.APL));
attrVals.Add(new KnownNumericalValue(feature.PV));
attrVals.Add(new KnownNumericalValue(feature.DPL));
attrVals.Add(new KnownNumericalValue(feature.PPS));
attrVals.Add(new KnownSymbolicValue((int)feature.Protocol));
attrVals.Add(new KnownNumericalValue(feature.FPS));
attrVals.Add(new KnownNumericalValue(feature.AB));
attrVals.Add(new KnownNumericalValue(feature.TBT));
attrVals.Add(new KnownNumericalValue(feature.BS));
attrVals.Add(new KnownNumericalValue(feature.PS));
attrVals.Add(new KnownNumericalValue(feature.NNP));
attrVals.Add(new KnownNumericalValue(feature.NSP));
attrVals.Add(new KnownNumericalValue(feature.PSP));
attrVals.Add(new KnownNumericalValue(feature.Duration));
attrVals.Add(new KnownNumericalValue(feature.AIT));
attrVals.Add(new KnownNumericalValue(feature.IOPR));
attrVals.Add(new KnownNumericalValue(feature.Reconnect));
attrVals.Add(new KnownSymbolicValue(feature.Type));
// System.Diagnostics.Debug.WriteLine("TrainingThread2");
// attrVals.Add(new ((DateTime)feature.DetectionTimeStamp));
Item it = new Item(attrVals.ToArray());
if (feature.Type > 0) // if the flow is not normal, count
{
if (!maliciousFlowCounter.TryGetValue(feature.Type, out value))
{
maliciousFlowCounter.Add(feature.Type, 1);
}
else
{
maliciousFlowCounter[feature.Type]++;
}
}
itemSet.Add(it);
}
foreach (int index in maliciousFlowCounter.Keys)
{
System.Diagnostics.Debug.WriteLine("Number of Malicious Flows for type: " + Flow2.GetFlowTypeName(index) + " is: " + maliciousFlowCounter[index].ToString());
}
SymbolicAttribute goalAttribute = attrSet.FindByName("Type") as SymbolicAttribute;
List <Attribute> testAttributes = new List <Attribute>();
testAttributes.Add(attrSet.FindByName("PX"));
testAttributes.Add(attrSet.FindByName("APL"));
testAttributes.Add(attrSet.FindByName("PV"));
testAttributes.Add(attrSet.FindByName("DPL"));
testAttributes.Add(attrSet.FindByName("PPS"));
testAttributes.Add(attrSet.FindByName("Protocol"));
testAttributes.Add(attrSet.FindByName("FPS"));
// testAttributes.Add(attrSet.FindByName("Type"));
testAttributes.Add(attrSet.FindByName("AB"));
testAttributes.Add(attrSet.FindByName("TBT"));
testAttributes.Add(attrSet.FindByName("BS"));
testAttributes.Add(attrSet.FindByName("PS"));
testAttributes.Add(attrSet.FindByName("NNP"));
testAttributes.Add(attrSet.FindByName("NSP"));
testAttributes.Add(attrSet.FindByName("PSP"));
testAttributes.Add(attrSet.FindByName("Duration"));
testAttributes.Add(attrSet.FindByName("AIT"));
testAttributes.Add(attrSet.FindByName("IOPR"));
testAttributes.Add(attrSet.FindByName("Reconnect"));
// System.Diagnostics.Debug.WriteLine("TrainingThread3");
SimpleDecisionTreeBuilder builder = new SimpleDecisionTreeBuilder( /// create tree hear!
new WeightedItemSet(itemSet),
new AttributeSet(testAttributes),
goalAttribute);
builder.ScoreThreshold = 0.0001d; // 0.0001 * itemSet.Size();
System.Diagnostics.Debug.WriteLine("DT ScoreThreshold is " + builder.ScoreThreshold.ToString());
LearningDecisionTree dt = builder.Build();
TestDecisionTree tdt = new TestDecisionTree(dt);
StoreDecisionTree(tdt);
}
catch (ThreadInterruptedException)
{
;
}
catch (ThreadAbortException)
{
;
}
catch (Exception ex)
{
MessageBox.Show(ex.ToString(), "Error");
}
finally
{
menuTraining.Text = Properties.Resources.StartTrainingText;
}
}