Signal modeling and classification using a robust latent space model based on t distributions

Chatzis, SP; Kosmopoulos, DI; Varvarigou, TA

dc.contributor.author	Chatzis, SP	en
dc.contributor.author	Kosmopoulos, DI	en
dc.contributor.author	Varvarigou, TA	en
dc.date.accessioned	2014-03-01T01:29:07Z
dc.date.available	2014-03-01T01:29:07Z
dc.date.issued	2008	en
dc.identifier.issn	1053-587X	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/19146
dc.subject	Bayesian inference	en
dc.subject	Latent subspace modeling	en
dc.subject	Pattern classification	en
dc.subject	Robust clustering methods	en
dc.subject.classification	Engineering, Electrical & Electronic	en
dc.subject.other	Bayesian networks	en
dc.subject.other	Cluster analysis	en
dc.subject.other	Covariance matrix	en
dc.subject.other	Normal distribution	en
dc.subject.other	Statistical methods	en
dc.subject.other	Variational techniques	en
dc.subject.other	Bayesian inference	en
dc.subject.other	Expectation-maximization algorithm	en
dc.subject.other	Latent subspace modeling	en
dc.subject.other	Classification (of information)	en
dc.title	Signal modeling and classification using a robust latent space model based on t distributions	en
heal.type	journalArticle	en
heal.identifier.primary	10.1109/TSP.2007.907912	en
heal.identifier.secondary	http://dx.doi.org/10.1109/TSP.2007.907912	en
heal.language	English	en
heal.publicationDate	2008	en
heal.abstract	Factor analysis is a statistical covariance modeling technique based on the assumption of normally distributed data. A mixture of factor analyzers can be hence viewed as a special case of Gaussian (normal) mixture models providing a mathematically sound framework for attribute space dimensionality reduction. A significant shortcoming of mixtures of factor analyzers is the vulnerability of normal distributions to outliers. Recently, the replacement of normal distributions with the heavier-tailed Student's-t distributions has been proposed as a way to mitigate these shortcomings and the treatment of the resulting model under an expectation-maximization (EM) algorithm framework has been conducted. In this paper, we develop a Bayesian approach to factor analysis modeling based on Student's-t distributions. We derive a tractable variational inference algorithm for this model by expressing the Student's-t distributed factor analyzers as a marginalization over additional latent variables. Our innovative approach provides an efficient and more robust alternative to EM-based methods, resolving their singularity and overfitting proneness problems, while allowing for the automatic determination of the optimal model size. We demonstrate the superiority of the proposed model over well-known covariance modeling techniques in a wide range of signal processing applications. © 2008 IEEE.	en
heal.publisher	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC	en
heal.journalName	IEEE Transactions on Signal Processing	en
dc.identifier.doi	10.1109/TSP.2007.907912	en
dc.identifier.isi	ISI:000253358400007	en
dc.identifier.volume	56	en
dc.identifier.issue	3	en
dc.identifier.spage	949	en
dc.identifier.epage	963	en