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

On feature selection and classificat...

Columbia University.

FindBook

Google Book

Amazon

博客來

On feature selection and classification in high dimensions.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	On feature selection and classification in high dimensions./
作者:	Ding, Yuejing.
面頁冊數:	149 p.
附註:	Adviser: Tian Zheng.
Contained By:	Dissertation Abstracts International69-05B.
標題:	Biology, Bioinformatics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3317545
ISBN:	9780549655602

On feature selection and classification in high dimensions.
Ding, Yuejing.

On feature selection and classification in high dimensions. - 149 p.

Adviser: Tian Zheng.

Thesis (Ph.D.)--Columbia University, 2008.

High dimensional data, such as that of gene expression, has provided vast amounts of information for scientific research and learning. However, in most cases, the information is much diluted by noises from non-informative features. Feature selection has become a necessary step for learning in high dimensions. It is also widely acknowledged that feature selection and classification methods for such data should consider possible interactions among the features, since they may carry stronger signals and reveal important scientific findings (such as gene-gene interactions).

ISBN: 9780549655602Subjects--Topical Terms:

1018415
Biology, Bioinformatics.

On feature selection and classification in high dimensions.
LDR:02707nam 2200313 a 45 001 852994
005 20100701
008 100701s2008 ||||||||||||||||| ||eng d
020 $a 9780549655602
035 $a (UMI)AAI3317545
035 $a AAI3317545
040 $a UMI $c UMI
100 1 $a Ding, Yuejing. $3 1019143
245 1 0 $a On feature selection and classification in high dimensions.
300 $a 149 p.
500 $a Adviser: Tian Zheng.
500 $a Source: Dissertation Abstracts International, Volume: 69-05, Section: B, page: 3073.
502 $a Thesis (Ph.D.)--Columbia University, 2008.
520 $a High dimensional data, such as that of gene expression, has provided vast amounts of information for scientific research and learning. However, in most cases, the information is much diluted by noises from non-informative features. Feature selection has become a necessary step for learning in high dimensions. It is also widely acknowledged that feature selection and classification methods for such data should consider possible interactions among the features, since they may carry stronger signals and reveal important scientific findings (such as gene-gene interactions).
520 $a In the first part of the thesis, we develop an associate score measuring the information in a feature subspace regarding to the class label, using local neighborhood patterns (k nearest neighbor patterns, KNNP). In the second part of the thesis, we propose a partition-based feature association score, the Sum of Squared Local Sums (SSLS). Both scores are model-free, with the advantage of considering higher order interaction among features while maintaining easy interpretability. Two backward elimination algorithms based on random subspaces are carried out to identify the best feature subspaces with top values of the two association scores respectively. Two classifiers using the selected subspaces based on these two scores are further proposed.
520 $a Through simulation and real data, both methods demonstrate power in identifying patterns informative about the class difference, not only in one dimensional subspace but also in higher dimensional subspaces. As a result, both methods outperform all SVM, Random Forests, Nearest Shrunken Centroid (NSC) and Golub's method in cancer classification. Moreover, both methods are nonparametric and can be applied to a wide variety of problems, not limited to gene expression data.
590 $a School code: 0054.
650 4 $a Biology, Bioinformatics. $3 1018415
650 4 $a Computer Science. $3 626642
650 4 $a Statistics. $3 517247
690 $a 0463
690 $a 0715
690 $a 0984
710 2 $a Columbia University. $3 571054
773 0 $t Dissertation Abstracts International $g 69-05B.
790 $a 0054
790 1 0 $a Zheng, Tian, $e advisor
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3317545