File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: Gaussian process dynamical models for human motion

TitleGaussian process dynamical models for human motion
Authors
Issue Date2008
Citation
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, v. 30 n. 2, p. 283-298 How to Cite?
AbstractWe introduce Gaussian process dynamical models (GPDM) for nonlinear time series analysis, with applications to learning models of human pose and motion from high-dimensionalmotion capture data. A GPDM is a latent variable model. It comprises a low-dimensional latent space with associated dynamics, and a map from the latent space to an observation space. We marginalize out the model parameters in closed-form, using Gaussian process priors for both the dynamics and the observation mappings. This results in a non-parametric model for dynamical systems that accounts for uncertainty in the model. We demonstrate the approach, and compare four learning algorithms on human motion capture data in which each pose is 50-dimensional. Despite the use of small data sets, the GPDM learns an effective representation of the nonlinear dynamics in these spaces. © 2008 IEEE.
Persistent Identifierhttp://hdl.handle.net/10722/192717
ISSN
2015 Impact Factor: 6.077
2015 SCImago Journal Rankings: 7.653
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorWang, JMen_US
dc.contributor.authorFleet, DJen_US
dc.contributor.authorHertzmann, Aen_US
dc.date.accessioned2013-11-20T04:58:59Z-
dc.date.available2013-11-20T04:58:59Z-
dc.date.issued2008en_US
dc.identifier.citationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, v. 30 n. 2, p. 283-298en_US
dc.identifier.issn0162-8828en_US
dc.identifier.urihttp://hdl.handle.net/10722/192717-
dc.description.abstractWe introduce Gaussian process dynamical models (GPDM) for nonlinear time series analysis, with applications to learning models of human pose and motion from high-dimensionalmotion capture data. A GPDM is a latent variable model. It comprises a low-dimensional latent space with associated dynamics, and a map from the latent space to an observation space. We marginalize out the model parameters in closed-form, using Gaussian process priors for both the dynamics and the observation mappings. This results in a non-parametric model for dynamical systems that accounts for uncertainty in the model. We demonstrate the approach, and compare four learning algorithms on human motion capture data in which each pose is 50-dimensional. Despite the use of small data sets, the GPDM learns an effective representation of the nonlinear dynamics in these spaces. © 2008 IEEE.en_US
dc.languageengen_US
dc.relation.ispartofIEEE Transactions on Pattern Analysis and Machine Intelligenceen_US
dc.titleGaussian process dynamical models for human motionen_US
dc.typeArticleen_US
dc.description.naturelink_to_subscribed_fulltexten_US
dc.identifier.doi10.1109/TPAMI.2007.1167en_US
dc.identifier.pmid18084059-
dc.identifier.scopuseid_2-s2.0-37549055132en_US
dc.identifier.volume30en_US
dc.identifier.issue2en_US
dc.identifier.spage283en_US
dc.identifier.epage298en_US
dc.identifier.isiWOS:000251580300007-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats