國立高雄大學圖資館 |

Language: English

Back

Simultaneous learning of non-linear ...

Boston University.

Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series.
Author:	Li, Rui.
Description:	123 p.
Notes:	Source: Dissertation Abstracts International, Volume: 70-10, Section: B, page: 6348.
Notes:	Adviser: Stan Sclaroff.
Contained By:	Dissertation Abstracts International70-10B.
Subject:	Computer Science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3382531
ISBN:	9781109435580

Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series.
Li, Rui.

Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series. - 123 p.

Source: Dissertation Abstracts International, Volume: 70-10, Section: B, page: 6348.

Thesis (Ph.D.)--Boston University, 2010.

The 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.

ISBN: 9781109435580Subjects--Topical Terms:

212513
Computer Science.

Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series.
LDR:02556nmm 2200265 4500 001 280740
005 20110119094942.5
008 110301s2010 ||||||||||||||||| ||eng d
020 $a 9781109435580
035 $a (UMI)AAI3382531
035 $a AAI3382531
040 $a UMI $c UMI
100 1 $a Li, Rui. $3 492822
245 1 0 $a Simultaneous learning of non-linear manifold and dynamical models for high-dimensional time series.
300 $a 123 p.
500 $a Source: Dissertation Abstracts International, Volume: 70-10, Section: B, page: 6348.
500 $a Adviser: Stan Sclaroff.
502 $a Thesis (Ph.D.)--Boston University, 2010.
520 $a The 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.
590 $a School code: 0017.
650 4 $a Computer Science. $3 212513
690 $a 0984
710 2 $a Boston University. $3 212722
773 0 $t Dissertation Abstracts International $g 70-10B.
790 1 0 $a Sclaroff, Stan, $e advisor
790 $a 0017
791 $a Ph.D.
792 $a 2010
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3382531