C ++ GMM Implementation Using Gibbs Sampler ie Dirichlet Gaussian Mixture Model

Question

I am looking for a C ++ implementation of multivariate GMM that uses the Gibbs Sampling approach to fit / classify (rather than regular EM) to be able to make full use of a priori information and add in constraints. Often known as the model of a Gaussian Dirichlet mixture or DPGMM.

I already implemented this in Matlab, but instead of wasting time converting this code (yes, I use the encoder built in Matlab to convert, but it currently uses various additional Matlab libraries). Efficiency is also important, I will install GMM on large data sets many times per second.

Thus, I am interested to know if there was already a well-known efficient code there. The initial search did not return very strongly.

Answer 1

Although this does not apply to GMM, you can use the CppBugs project to specify your own model and let the library run the simulation.