| Paper: | MLSP-P1.3 | ||
| Session: | Blind Source Separation II | ||
| Time: | Tuesday, May 16, 14:00 - 16:00 | ||
| Presentation: | Poster | ||
| Topic: | Machine Learning for Signal Processing: Blind Signal Separation and Independent Component Analysis | ||
| Title: | A Gibbs Sampling Approach to Independent Factor Analysis | ||
| Authors: | Omolabake A. Adenle, William J. Fitzgerald, University of Cambridge, United Kingdom | ||
| Abstract: | We present a Gibbs sampler for estimating parameters of the Independent Factor model. Independent Factor Analysis (IFA) is a method for learning local subspaces in data e.g., Mixtures of Factor Analyzers. It can also be considered a method for blind source separation. The generative model for IFA is hierarchical and each factor is modeled as an independent Mixture of Gaussians, each mixture component representing a factor state. Computing expectations over factors quickly becomes intractable with increasing number of factors as this requires summation over exponentially many state configurations, making parameter estimation via EM infeasible. Unlike the Variational method that has been proposed, we take a simulation based approach to obtain exact parameter estimates. We define prior distributions and use a Gibbs sampler to obtain samples from the parameter posterior. Application to synthetic data demonstrates effectiveness of the method in estimating model parameters and robustness to model permutation invariance. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012