東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Multiple model estimation using SVM-...

Ma, Yunqian.

Linked to FindBook

Google Book

Amazon

博客來

Multiple model estimation using SVM-based learning.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Multiple model estimation using SVM-based learning./
Author:	Ma, Yunqian.
Description:	152 p.
Notes:	Source: Dissertation Abstracts International, Volume: 64-10, Section: B, page: 5131.
Contained By:	Dissertation Abstracts International64-10B.
Subject:	Engineering, Electronics and Electrical. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3109360
ISBN:	0496569023

Multiple model estimation using SVM-based learning.
Ma, Yunqian.

Multiple model estimation using SVM-based learning. - 152 p.

Source: Dissertation Abstracts International, Volume: 64-10, Section: B, page: 5131.

Thesis (Ph.D.)--University of Minnesota, 2003.

Most existing learning algorithms have been developed for standard formulations of the learning problem, such as classification or regression. These standard inductive learning formulations assume that all available (or training) data can be well described by a single statistical model. For example, in classification setting the goal is to estimate a single decision boundary (which may be complex or nonlinear). Likewise, under regression formulation, the goal is to estimate a single target function from finite and noisy data samples. However, there are many applications which naturally lead to learning formulations where the goal is to estimate several models from available (finite) data. For such applications, it may be reasonable to assume that the data is generated by several (unknown) models; however we do not know which model has generated a particular sample. For example, in computer vision, a sequence of video frames may contain several moving objects (with overlapping trajectories), so the problem of multiple motion estimation can be viewed under multiple model estimation framework. For such multiple model estimation problems, the goal of learning is to separate available data into several (disjoint) subsets, and at the same time estimate unknown models corresponding to each subset of data. This project describes several useful multiple model formulations, such as multiple model classification, multiple model regression, and multiple model clustering, and several real-life application settings where multiple model formulation approach can be naturally applied.

ISBN: 0496569023Subjects--Topical Terms:

626636
Engineering, Electronics and Electrical.

Multiple model estimation using SVM-based learning.
LDR:03343nmm 2200301 4500 001 1847508
005 20051108095419.5
008 130614s2003 eng d
020 $a 0496569023
035 $a (UnM)AAI3109360
035 $a AAI3109360
040 $a UnM $c UnM
100 1 $a Ma, Yunqian. $3 1568445
245 1 0 $a Multiple model estimation using SVM-based learning.
300 $a 152 p.
500 $a Source: Dissertation Abstracts International, Volume: 64-10, Section: B, page: 5131.
500 $a Adviser: Vladimir S. Cherkassky.
502 $a Thesis (Ph.D.)--University of Minnesota, 2003.
520 $a Most existing learning algorithms have been developed for standard formulations of the learning problem, such as classification or regression. These standard inductive learning formulations assume that all available (or training) data can be well described by a single statistical model. For example, in classification setting the goal is to estimate a single decision boundary (which may be complex or nonlinear). Likewise, under regression formulation, the goal is to estimate a single target function from finite and noisy data samples. However, there are many applications which naturally lead to learning formulations where the goal is to estimate several models from available (finite) data. For such applications, it may be reasonable to assume that the data is generated by several (unknown) models; however we do not know which model has generated a particular sample. For example, in computer vision, a sequence of video frames may contain several moving objects (with overlapping trajectories), so the problem of multiple motion estimation can be viewed under multiple model estimation framework. For such multiple model estimation problems, the goal of learning is to separate available data into several (disjoint) subsets, and at the same time estimate unknown models corresponding to each subset of data. This project describes several useful multiple model formulations, such as multiple model classification, multiple model regression, and multiple model clustering, and several real-life application settings where multiple model formulation approach can be naturally applied.
520 $a We also describe a class of constructive learning algorithms for multiple model estimation. The proposed model estimation strategy assumes that the majority of available data is generated by a single model; hence we use robust methodology to estimate this (major) model. We use Support Vector Machine (SVM) based algorithm for robust model estimation. We developed SVM-based algorithms for multiple model regression and multiple model classification. Empirical comparisons performed using several simulated and real-life data sets indicate superior performance of the proposed learning algorithms.
520 $a Finally, this thesis presents new methodology for analytic setting of hyper-parameters for SVM regression. This is important for practical applications of SVM regression in general, and for multiple model regression estimation, in particular.
590 $a School code: 0130.
650 4 $a Engineering, Electronics and Electrical. $3 626636
650 4 $a Artificial Intelligence. $3 769149
690 $a 0544
690 $a 0800
710 2 0 $a University of Minnesota. $3 676231
773 0 $t Dissertation Abstracts International $g 64-10B.
790 1 0 $a Cherkassky, Vladimir S., $e advisor
790 $a 0130
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3109360