Parameter Estimation Of Gaussian Mixture Model Based On Variational Bayesian Learning

Parameter estimation of gaussian mixture model has recently seen significant activity in many areas of science and engineering, motivated by recent theoretical results for researching on global optimal convergence and interesting practical application- s. In this paper, we present novel evaluation algorithm for estimating the parameter of gaussian mixture model based on Variational Bayesian Learning (VBL) principles. Starting from a traditional VBEM algorithm and analysing the convergence of the algorithm as an initial-value constraint, we develop an approach that is very effective in estimating the parameter of gaussian mixture model while providing high astringency performance. We provide empirical results and comparison with current VBEM algorithm that illustrate the effectiveness of this approach.