東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Learning from data locally and globally.

Huang, Kaizhu.

FindBook

Google Book

Amazon

博客來

Learning from data locally and globally.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Learning from data locally and globally./
作者:	Huang, Kaizhu.
面頁冊數:	194 p.
附註:	Source: Dissertation Abstracts International, Volume: 66-07, Section: B, page: 3794.
Contained By:	Dissertation Abstracts International66-07B.
標題:	Computer Science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3182146
ISBN:	0542235307

Learning from data locally and globally.
Huang, Kaizhu.

Learning from data locally and globally. - 194 p.

Source: Dissertation Abstracts International, Volume: 66-07, Section: B, page: 3794.

Thesis (Ph.D.)--The Chinese University of Hong Kong (People's Republic of China), 2004.

I mainly consider the task of learning classifiers from data in this thesis. In this context, I propose a common framework that combines two different and important paradigms in machine learning: global learning and local learning. Traditional global learning approaches focus on describing phenomena by attempting to estimate a distribution from data. Based on the estimated distribution, the global learning methods can then perform inferences, conduct marginalizations, and make predictions. Although enjoying a long and distinguished history and containing many good features, e.g., a relatively simple optimization, and the flexibility in incorporating global information such as structure information and invariance etc., these learning approaches usually have to assume a specific type of distribution a prior. Therefore, they are widely argued for lacking the generality. On the other hand, local learning methods do not estimate a distribution from data. Instead, they focus on extracting only the local information, which is directly related to the learning task, i.e., the classification in this thesis. Recent progress following this trend has demonstrated that local learning approaches, e.g., Support Vector Machines (SVM), outperform the global learning methods in many aspects. Despite of the success, local learning approaches actually discard plenty of important global information on data, e.g., the structure information. Therefore, this restricts the learning performance of this types of learning schemes.

ISBN: 0542235307Subjects--Topical Terms:

626642
Computer Science.

Learning from data locally and globally.
LDR:03378nmm 2200289 4500 001 1819221
005 20061004161535.5
008 130610s2004 eng d
020 $a 0542235307
035 $a (UnM)AAI3182146
035 $a AAI3182146
040 $a UnM $c UnM
100 1 $a Huang, Kaizhu. $3 923774
245 1 0 $a Learning from data locally and globally.
300 $a 194 p.
500 $a Source: Dissertation Abstracts International, Volume: 66-07, Section: B, page: 3794.
500 $a Advisers: King Kuo Chin; Lyu Rung Tsong.
502 $a Thesis (Ph.D.)--The Chinese University of Hong Kong (People's Republic of China), 2004.
520 $a I mainly consider the task of learning classifiers from data in this thesis. In this context, I propose a common framework that combines two different and important paradigms in machine learning: global learning and local learning. Traditional global learning approaches focus on describing phenomena by attempting to estimate a distribution from data. Based on the estimated distribution, the global learning methods can then perform inferences, conduct marginalizations, and make predictions. Although enjoying a long and distinguished history and containing many good features, e.g., a relatively simple optimization, and the flexibility in incorporating global information such as structure information and invariance etc., these learning approaches usually have to assume a specific type of distribution a prior. Therefore, they are widely argued for lacking the generality. On the other hand, local learning methods do not estimate a distribution from data. Instead, they focus on extracting only the local information, which is directly related to the learning task, i.e., the classification in this thesis. Recent progress following this trend has demonstrated that local learning approaches, e.g., Support Vector Machines (SVM), outperform the global learning methods in many aspects. Despite of the success, local learning approaches actually discard plenty of important global information on data, e.g., the structure information. Therefore, this restricts the learning performance of this types of learning schemes.
520 $a In this thesis, I thus develop a hybrid model named Maxi-Min Margin Machine (M4), which successfully combines two largely differently but complementary paradigms. Within this new framework, I propose a hybrid model named Maxi-Min Margin Machine (M4). This model is demonstrated to contain both appealing features in global learning and local learning. It not only captures the global structure information from data, but it also provides a task-oriented scheme for the learning purpose and inherits the superior performance from local learning. As a major contribution, M 4 successfully unifies many important learning models, including Support Vector Machines, Minimax Probability Machine (MPM), and Fisher Discriminant Analysis. Another compelling feature of M4 is that it can be cast as a Sequential Second Order Cone Programming problem, yielding a polynomial time complexity. (Abstract shortened by UMI.)
590 $a School code: 1307.
650 4 $a Computer Science. $3 626642
690 $a 0984
710 2 0 $a The Chinese University of Hong Kong (People's Republic of China). $3 1265600
773 0 $t Dissertation Abstracts International $g 66-07B.
790 1 0 $a Chin, King Kuo, $e advisor
790 1 0 $a Tsong, Lyu Rung, $e advisor
790 $a 1307
791 $a Ph.D.
792 $a 2004
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3182146