Variational Gaussian mixtures for blind source detection
نویسندگان
چکیده
Bayesian algorithms have lately been used in a large variety of applications. This paper proposes a new methodology for hyperparameter initialization in the Variational Bayes (VB) algorithm. We employ a dual expectationmaximization (EM) algorithm as the initialization stage in the VB-based learning. In the first stage, the EM algorithm is used on the given data set while the second EM algorithm is applied on distributions of parameters resulted from several runs of the first stage EM. The graphical model case study considered in this paper consists of a mixture of Gaussians. Appropriate conjugate prior distributions are considered for modelling the parameters. The proposed methodology is applied on blind source separation of modulated signals.
منابع مشابه
Nonlinear Blind Source Separation by Variational Bayesian Learning
Blind separation of sources from their linear mixtures is a well understood problem. However, if the mixtures are nonlinear, this problem becomes generally very difficult. This is because both the nonlinear mapping and the underlying sources must be learned from the data in a blind manner, and the problem is highly ill-posed without a suitable regularization. In our approach, multilayer percept...
متن کاملUnderdetermined Model-Based Blind Source Separation of Reverberant Speech Mixtures using Spatial Cues in a Variational Bayesian Framework
In this paper, we propose a new method for underdetermined blind source separation of reverberant speech mixtures by classifying each time-frequency (T-F) point of the mixtures according to a combined variational Bayesian model of spatial cues, under sparse signal representation assumption. We model the T-F observations by a variational mixture of circularly-symmetric complex-Gaussians. The spa...
متن کاملA Neural Network for the Blind Separation of Non-Gaussian Sources
| In this paper, a two{layer neural network is presented that organizes itself to perform blind source separation. The inputs to the network are prewhitened linear mixtures of unknown independent source signals. An unsu-pervised nonlinear hebbian learning rule is used for training the network. After convergence, the network is able to extract the source signals from the mixtures, provided that ...
متن کاملState space blind source recovery for mixtures of multiple source distributions
The paper discusses State Space Blind Source Recovery (BSR) for minimum phase and non-minimum phase mixtures of gaussian and non-gaussian distributions. The State Space Natural Gradient approach results in compact iterative update laws for BSR. Two separate state space algorithms for minimum phase and non-minimum phase mixing environments are presented. The advantages and disadvantages of both ...
متن کاملWavelet De-noising for Blind Source Separation in Noisy Mixtures
Blind source separation, which supposes that the sources are independent, is a well known domain in signal processing. However, in a noisy environment the estimation of the criterion is harder due to the noise. In strong noisy mixtures, we propose two new principles based on the combination of wavelet de-noising processing and blind source separation. We compare them in the cases of white/corre...
متن کامل