public void Learn(List <DataElement> data)
{
kMeans _kMeans = new kMeans(d);
_gaussianDistributions = _kMeans.GetGroupRepresentatives(data, threshold);
List <double[][]> mfccVectors = new List <double[][]>();
for (int i = 0; i < data.Count; i++)
{
mfccVectors.Add(data[i].MfccCoefficients);
}
em = new ExpectationMaximization(_gaussianDistributions.ToArray(), mfccVectors);
int iterations = 0;
while (iterations < maxIterations)
{
List <DataElement> prevMeans = new List <DataElement>();
foreach (var gd in _gaussianDistributions)
{
prevMeans.Add(new DataElement(gd.Mean));
}
em.Compute();
List <DataElement> currMeans = new List <DataElement>();
foreach (var gd in _gaussianDistributions)
{
currMeans.Add(new DataElement(gd.Mean));
}
List <double> meansDistances = MathHelper.CountMeansDistances(prevMeans, currMeans);
if (!MathHelper.meansDistancesChanged(threshold, meansDistances))
{
break;
}
iterations++;
}
AfterLearning = true;
}
public GMM()
{
Name = "GMM";
kMeans _kMeans = new kMeans(Enum.GetNames(typeof(SongGenre)).Length);
//_gaussianDistributions = _kMeans.GetGroupRepresentatives(TrainingData);
}