Newbie question: Extract mean and variance from a gaussian(mean, variance)

• terça-feira, 19 de junho de 2012 08:49

   

   
  

  0

  Goood Morning Infer.NET team and thanks for work on this wonderful library,

  As a newbie in Infer.net, working in VB.NET, i am trying to test bayesian inference.

  All is ok, i can create a w matrix, calculated with by this way:

  Dim y_observed As VariableArray(Of Double) = Variable.Observed(Variation, r).Named("y")
  Dim x_observed As VariableArray(Of Vector) = Variable.Observed(x, r).Named("x")

  ' Bayes Point Machine
  Dim noise As Double = 0.1
  y_observed(r) = Variable.GaussianFromMeanAndVariance(Variable.InnerProduct(w, x_observed(r)), noise)

  I "infer" my wposterior gaussian like that: Dim wPosterior As VectorGaussian = engine.Infer(Of VectorGaussian)(w)

  and i use it with a unique  "Out of the Sample" observation  yOTS(0) where:  

  Dim yOTS As VariableArray(Of Double) = Variable.Array(Of Double)(r2)

  yOTS(0) = Variable.GaussianFromMeanAndVariance(Variable.InnerProduct(Variable.Random(wPosterior), xOTS_observed(0)), noise)

  Finally i obtain my output, for exemple:

  "output=" & engine.Infer(yOTS).ToString => output=[0] Gaussian(0.00123, 0.1196)

  My problem is quite frustrating: i can t put mean (here 0.00123) and variance (0.1196) in variables ...

  I have tried moyenne = engine.Infer(yOTS).GetMean(), it should be straightforward but ... i have a bug, the error message says:

  "The public member 'GetMean' from type 'DistributionStrucArray(Of Gaussain, Double)' is not found' (Rough english translation from french error message).

  So, is there a simplier way to get the mean and variance, what have i to do to get these values?

  Thanks for help.

  • Responder
  • Citação