private void LearnClick5LabelModel(
int numLabels,
bool learnScoreMean,
bool learnScorePrec,
bool learnJudgePrec,
bool learnClickPrec,
bool learnThresholds,
double nominalScoreMean,
double nominalScorePrec,
double nominalJudgePrec,
double nominalClickPrec,
int[] labels,
int[] clicks,
int[] exams,
int chunkSize,
int nPasses,
bool printToConsole,
out Gaussian margScoreMean,
out Gamma margScorePrec,
out Gamma margJudgePrec,
out Gamma margClickPrec,
out Gaussian[] margThresh)
{
InferenceEngine engine = new InferenceEngine();
int numThresholds = numLabels + 1; // Includes end-points
//-------------------------------------------------------------
// Priors
//-------------------------------------------------------------
Gaussian priorScoreMean = Gaussian.FromMeanAndVariance(nominalScoreMean, learnScoreMean ? 1 : 0);
Gamma priorScorePrec = Gamma.FromMeanAndVariance(nominalScorePrec, learnScorePrec ? 1 : 0);
Gamma priorJudgePrec = Gamma.FromMeanAndVariance(nominalJudgePrec, learnJudgePrec ? 1 : 0);
Gamma priorClickPrec = Gamma.FromMeanAndVariance(nominalClickPrec, learnClickPrec ? 1 : 0);
Gaussian[] priorThreshMean;
CalculatePriors(learnThresholds, numLabels, out priorThreshMean);
//------------------------------------------------------
// Observations
//------------------------------------------------------
Gaussian[][][] clickObs = getClickObservations(numLabels, chunkSize, labels, clicks, exams);
int numChunks = clickObs.Length;
//-----------------------------------------------------
// Create an array of batch variables
//-----------------------------------------------------
Model model = new Model(numChunks);
SharedVariable <double, Gaussian> scoreMean = (SharedVariable <double, Gaussian>) SharedVariable <double> .Random(priorScoreMean).Named("scoreMean");
SharedVariable <double, Gamma> scorePrec = (SharedVariable <double, Gamma>) SharedVariable <double> .Random(priorScorePrec).Named("scorePrec");
SharedVariable <double, Gamma> judgePrec = (SharedVariable <double, Gamma>) SharedVariable <double> .Random(priorJudgePrec).Named("judgePrec");
SharedVariable <double, Gamma> clickPrec = (SharedVariable <double, Gamma>) SharedVariable <double> .Random(priorClickPrec).Named("clickPrec");
SharedVariable <double, Gaussian>[] threshMeans = new SharedVariable <double, Gaussian> [numThresholds];
for (int t = 0; t < numThresholds; t++)
{
threshMeans[t] = (SharedVariable <double, Gaussian>) SharedVariable <double, Gaussian> .Random(priorThreshMean[t]).Named("threshMeans" + t);
}
engine.Compiler.DeclarationProvider = Microsoft.ML.Probabilistic.Compiler.RoslynDeclarationProvider.Instance;
var info = GetType().GetMethod("Click5LabelModel", System.Reflection.BindingFlags.NonPublic | System.Reflection.BindingFlags.Instance);
var ca = engine.Compiler.CompileWithoutParams(info);
//----------------------------------------------------------
// Outer loop iterates over a number of passes
// Inner loop iterates over the unique labels
//----------------------------------------------------------
for (int pass = 0; pass < nPasses; pass++)
{
for (int c = 0; c < numChunks; c++)
{
Gaussian[] threshInputs = new Gaussian[threshMeans.Length];
for (int i = 0; i < threshInputs.Length; i++)
{
threshInputs[i] = ((SharedVariable <double, Gaussian>)threshMeans[i]).MessageToBatch(model, c);
}
ca.SetObservedValue("inputScoreMean", scoreMean.MessageToBatch(model, c));
ca.SetObservedValue("inputScorePrec", scorePrec.MessageToBatch(model, c));
ca.SetObservedValue("inputJudgePrec", judgePrec.MessageToBatch(model, c));
ca.SetObservedValue("inputClickPrec", clickPrec.MessageToBatch(model, c));
ca.SetObservedValue("inputThresh", threshInputs);
ca.SetObservedValue("clickObservations", clickObs[c]);
ca.Execute(10);
// Retrieve the output messages
model.InferShared(ca, c);
if (printToConsole)
{
Console.WriteLine("****** Pass {0}, chunk {1} ******", pass, c);
Console.WriteLine("----- Marginals -----");
Console.WriteLine("scoreMean = {0}", scoreMean.Marginal <Gaussian>());
Console.WriteLine("scorePrec = {0}", scorePrec.Marginal <Gamma>());
Console.WriteLine("judgePrec = {0}", judgePrec.Marginal <Gamma>());
Console.WriteLine("clickPrec = {0}", clickPrec.Marginal <Gamma>());
for (int t = 0; t < numThresholds; t++)
{
Console.WriteLine("threshMean {0} = {1}", t, threshMeans[t].Marginal <Gaussian>());
}
}
}
}
margScoreMean = scoreMean.Marginal <Gaussian>();
margScorePrec = scorePrec.Marginal <Gamma>();
margJudgePrec = judgePrec.Marginal <Gamma>();
margClickPrec = clickPrec.Marginal <Gamma>();
margThresh = new Gaussian[numThresholds];
for (int i = 0; i < margThresh.Length; i++)
{
margThresh[i] = threshMeans[i].Marginal <Gaussian>();
}
}