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

Improving the classification of micr...

Liu, Shuang.

FindBook

Google Book

Amazon

博客來

Improving the classification of microarray data: Supervised and unsupervised methods.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Improving the classification of microarray data: Supervised and unsupervised methods./
作者:	Liu, Shuang.
面頁冊數:	96 p.
附註:	Source: Dissertation Abstracts International, Volume: 69-06, Section: B, page: 3613.
Contained By:	Dissertation Abstracts International69-06B.
標題:	Mathematics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3317946
ISBN:	9780549671725

Improving the classification of microarray data: Supervised and unsupervised methods.
Liu, Shuang.

Improving the classification of microarray data: Supervised and unsupervised methods. - 96 p.

Source: Dissertation Abstracts International, Volume: 69-06, Section: B, page: 3613.

Thesis (Ph.D.)--University of California, Davis, 2008.

Due to the high dimensionality and significant noise in microarray data, the differences between gene groups are not well explained by many classification approaches. In this study of differentially regulated gene groups and pathways, improved procedures for supervised and unsupervised two-group classification procedures were proposed and evaluated. Meanwhile, we examined the effect of the gene selection steps and the dimension reduction methods in the classification of the differentially treated samples from the mitochondrial DNA study.

ISBN: 9780549671725Subjects--Topical Terms:

515831
Mathematics.

Improving the classification of microarray data: Supervised and unsupervised methods.
LDR:03381nam 2200301 4500 001 1391201
005 20110110132552.5
008 130515s2008 ||||||||||||||||| ||eng d
020 $a 9780549671725
035 $a (UMI)AAI3317946
035 $a AAI3317946
040 $a UMI $c UMI
100 1 $a Liu, Shuang. $3 1669598
245 1 0 $a Improving the classification of microarray data: Supervised and unsupervised methods.
300 $a 96 p.
500 $a Source: Dissertation Abstracts International, Volume: 69-06, Section: B, page: 3613.
502 $a Thesis (Ph.D.)--University of California, Davis, 2008.
520 $a Due to the high dimensionality and significant noise in microarray data, the differences between gene groups are not well explained by many classification approaches. In this study of differentially regulated gene groups and pathways, improved procedures for supervised and unsupervised two-group classification procedures were proposed and evaluated. Meanwhile, we examined the effect of the gene selection steps and the dimension reduction methods in the classification of the differentially treated samples from the mitochondrial DNA study.
520 $a Five supervised classification methods, including Mahalanobis Distance classifier, Logistic Regression, Support Vector Machine, and two forms of K-nearest neighbor classifier, were applied to classify gene groups. Misclassification rates were then estimated using the leave-one-out cross-validation for each two-group classification case. Finally, a self-designed permutation test was used to evaluate the classification results by comparing the grouped misclassification rates with respect to the study purposes. In this way, we were able to reveal the biological differences between gene groups and compare the performance of the classification methods statistically. The permutation test showed that the K-nearest neighbor method based on the correlation distance appeared to be a solution to the problem of high dimensionality.
520 $a As for the unsupervised classification method, we investigated the initialization based on the robust estimator and principal component analysis as a data transformation in the model-based clustering. The combination of these techniques improved the accuracy and efficiency in clustering high volume and high dimensional gene data. In addition, the results strongly supported initializing the EM algorithm using the Minimum Covariant Determinant estimator based on small sized samples.
520 $a The same five supervised classification processes were applied to classify the differentially treated samples from the mitochondrial DNA study. First, a subset of genes significantly responsive to the designed treatments across the samples was selected while controlling the false discovery rate. Then, the principal components based on the selected genes were used to summarize the variations between groups of samples. A permutation test was also applied to ensure the significance of the gene selection process. The test results showed the advantage of the k-nearest neighbors classifier when the number of samples was limited.
590 $a School code: 0029.
650 4 $a Mathematics. $3 515831
650 4 $a Statistics. $3 517247
650 4 $a Biology, Bioinformatics. $3 1018415
690 $a 0405
690 $a 0463
690 $a 0715
710 2 $a University of California, Davis. $3 1018682
773 0 $t Dissertation Abstracts International $g 69-06B.
790 $a 0029
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3317946