This is a python implementation for Variational Bayesian Learning.
Currently Gaussian Mixture Model (VBGMM) and Hidden Markov Model (VBHMM) are supported.
Conventional Expectation Maximization learning is also implemented for both GMM and HMM.
Non-parametric approach based on Stick Braking Process was included for alternative GMM.
Forward-Backward routines in HMM are accelerated by Fortran 90 with f2py.
Numpy and Scipy is needed.
C or Fortran compiler is also needed to use extension module of Forward-Backward.