Robust sequential data modeling using an outlier tolerant hidden markov model

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	2014-03-01T01:31:49Z
dc.date.available	2014-03-01T01:31:49Z
dc.date.issued	2009	en
dc.identifier.issn	0162-8828	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/19941
dc.subject	Expectation-maximization	en
dc.subject	Factor analysis	en
dc.subject	Hidden Markov models	en
dc.subject	Sequential data modeling	en
dc.subject	Student's t-distribution	en
dc.subject.classification	Computer Science, Artificial Intelligence	en
dc.subject.classification	Engineering, Electrical & Electronic	en
dc.subject.other	Conventional approach	en
dc.subject.other	Covariance matrices	en
dc.subject.other	Data sets	en
dc.subject.other	Expectation-maximization	en
dc.subject.other	Factor analysis	en
dc.subject.other	Finite gaussian mixture models	en
dc.subject.other	Finite mixtures	en
dc.subject.other	Gaussian Mixture Model	en
dc.subject.other	Hidden state	en
dc.subject.other	Model parameters estimation	en
dc.subject.other	Multivariate Student	en
dc.subject.other	Sequential data	en
dc.subject.other	Sequential data modeling	en
dc.subject.other	T-mixture models	en
dc.subject.other	Blind source separation	en
dc.subject.other	Communication channels (information theory)	en
dc.subject.other	Computational grammars	en
dc.subject.other	Covariance matrix	en
dc.subject.other	Maximum likelihood estimation	en
dc.subject.other	Maximum principle	en
dc.subject.other	Mixtures	en
dc.subject.other	Object recognition	en
dc.subject.other	Parameter estimation	en
dc.subject.other	Students	en
dc.subject.other	Hidden Markov models	en
dc.title	Robust sequential data modeling using an outlier tolerant hidden markov model	en
heal.type	journalArticle	en
heal.identifier.primary	10.1109/TPAMI.2008.215	en
heal.identifier.secondary	http://dx.doi.org/10.1109/TPAMI.2008.215	en
heal.language	English	en
heal.publicationDate	2009	en
heal.abstract	Hidden Markov (chain) models using finite Gaussian mixture models as their hidden state distributions have been successfully applied in sequential data modeling and classification applications. Nevertheless, Gaussian mixture models are well known to be highly intolerant to the presence of untypical data within the fitting data sets used for their estimation. Finite Student's t-mixture models have recently emerged as a heavier-tailed, robust alternative to Gaussian mixture models, overcoming these hurdles. To exploit these merits of Student's t-mixture models in the context of a sequential data modeling setting, we introduce, in this paper, a novel hidden Markov model where the hidden state distributions are considered to be finite mixtures of multivariate Student's t-densities. We derive an algorithm for the model parameters estimation under a maximum likelihood framework, assuming full, diagonal, and factor-analyzed covariance matrices. The advantages of the proposed model over conventional approaches are experimentally demonstrated through a series of sequential data modeling applications. © 2009 IEEE.	en
heal.publisher	IEEE COMPUTER SOC	en
heal.journalName	IEEE Transactions on Pattern Analysis and Machine Intelligence	en
dc.identifier.doi	10.1109/TPAMI.2008.215	en
dc.identifier.isi	ISI:000267369800009	en
dc.identifier.volume	31	en
dc.identifier.issue	9	en
dc.identifier.spage	1657	en
dc.identifier.epage	1669	en