Show simple item record

dc.contributor.authorLi, Ruien_US
dc.date.accessioned2011-10-20T04:58:51Z
dc.date.available2011-10-20T04:58:51Z
dc.date.issued2009-08-10en_US
dc.identifier.citationLi, Rui. "Simultaneous Learning Of Non-Linear Manifold And Dynamical Models For High-Dimensional Time Series (PhD Thesis)", Technical Report BUCS-TR-2009-027, Computer Science Department, Boston University, August 10, 2009. [Available from: http://hdl.handle.net/2144/1751]en_US
dc.identifier.urihttp://hdl.handle.net/2144/1751
dc.description.abstractThe goal of this work is to learn a parsimonious and informative representation for high-dimensional time series. Conceptually, this comprises two distinct yet tightly coupled tasks: learning a low-dimensional manifold and modeling the dynamical process. These two tasks have a complementary relationship as the temporal constraints provide valuable neighborhood information for dimensionality reduction and conversely, the low-dimensional space allows dynamics to be learnt efficiently. Solving these two tasks simultaneously allows important information to be exchanged mutually. If nonlinear models are required to capture the rich complexity of time series, then the learning problem becomes harder as the nonlinearities in both tasks are coupled. The proposed solution approximates the nonlinear manifold and dynamics using piecewise linear models. The interactions among the linear models are captured in a graphical model. The model structure setup and parameter learning are done using a variational Bayesian approach, which enables automatic Bayesian model structure selection, hence solving the problem of over-fitting. By exploiting the model structure, efficient inference and learning algorithms are obtained without oversimplifying the model of the underlying dynamical process. Evaluation of the proposed framework with competing approaches is conducted in three sets of experiments: dimensionality reduction and reconstruction using synthetic time series, video synthesis using a dynamic texture database, and human motion synthesis, classification and tracking on a benchmark data set. In all experiments, the proposed approach provides superior performance.en_US
dc.description.sponsorshipNational Science Foundation (IIS-0713168, CNS-0202067, IIS-0308213, IS-0208876)en_US
dc.language.isoen_USen_US
dc.publisherBoston University Computer Science Departmenten_US
dc.relation.ispartofseriesBUCS Technical Reports;BUCS-TR-2009-027en_US
dc.titleSimultaneous Learning of Non-Linear Manifold and Dynamical Models for High-Dimensional Time Seriesen_US
dc.typeTechnical Reporten_US
etd.degree.nameDoctor of Philosophy
etd.degree.leveldoctoral
etd.degree.disciplineComputer Science
etd.degree.grantorBoston University
dc.relation.isnodouble1445*


Files in this item

This item appears in the following Collection(s)

Show simple item record