public object Clone()
{
OneClassSVM clone = new OneClassSVM();
clone.copy(this);
return(clone);
}
public void copy(OneClassSVM rhs2)
{
param = rhs2.param == null ? null : (SVMParam)rhs2.param.Clone();
cross_validation = rhs2.cross_validation;
quiet = rhs2.quiet;
_threshold = rhs2._threshold;
model = rhs2.model == null ? null : (SVMModel)rhs2.model.Clone();
if (model != null)
{
model.Param = param;
}
}
public static void SolveOutlierDetectionProblem()
{
List <KeyValuePair <double[], double> > trainingBatch = new List <KeyValuePair <double[], double> >();
// add some normal data
for (int i = 0; i < 100; ++i)
{
trainingBatch.Add(new KeyValuePair <double[], double>(new double[] { randn() * 0.3 + 2, randn() * 0.3 + 2 }, -1.0));
trainingBatch.Add(new KeyValuePair <double[], double>(new double[] { randn() * 0.3 - 2, randn() * 0.3 - 2 }, -1.0));
}
List <KeyValuePair <double[], double> > crossValidationBatch = new List <KeyValuePair <double[], double> >();
// add some validation data
for (int i = 0; i < 20; ++i)
{
crossValidationBatch.Add(new KeyValuePair <double[], double>(new double[] { randn() * 0.3 + 2, randn() * 0.3 + 2 }, -1.0));
crossValidationBatch.Add(new KeyValuePair <double[], double>(new double[] { randn() * 0.3 - 2, randn() * 0.3 - 2 }, -1.0));
}
List <KeyValuePair <double[], double> > outliers = new List <KeyValuePair <double[], double> >();
// add some outliers data
for (int i = 0; i < 20; ++i)
{
outliers.Add(new KeyValuePair <double[], double>(new double[] { rand(-4, 4), rand(-4, 4) }, 1.0));
outliers.Add(new KeyValuePair <double[], double>(new double[] { rand(-4, 4), rand(-4, 4) }, 1.0));
}
OneClassSVM algorithm = new OneClassSVM();
algorithm.set_gamma(0.1);
algorithm.set_nu(0.1);
var batches = from p in trainingBatch
select p.Key;
algorithm.Fit(batches.ToList());
for (int i = 0; i < crossValidationBatch.Count; ++i)
{
double predicted = algorithm.IsOutlier(crossValidationBatch[i].Key) ? 1.0 : -1.0;
Console.WriteLine("predicted: " + predicted + "\texpected: " + crossValidationBatch[i].Value);
}
for (int i = 0; i < outliers.Count; ++i)
{
double predicted = algorithm.IsOutlier(crossValidationBatch[i].Key) ? 1.0 : -1.0;
Console.WriteLine("predicted: " + predicted + "\texpected: " + outliers[i].Value);
}
}
