DocumentCode
2541109
Title
Feature selection based on bayes minimum error probability
Author
Li, Jian ; Wei, Jin-Mao ; Yu, Tian ; Zhang, Hai-Wei
Author_Institution
Coll. of Inf. Tech. Sci., Nankai Univ., Tianjin, China
fYear
2012
fDate
29-31 May 2012
Firstpage
706
Lastpage
710
Abstract
Feature selection is very important to classification. In this paper, we propose to select features based on Bayes minimum error probability (SFBMEP). And we exploit the proposed method to sift possible functional genes for classifying cancers. The method dynamically evaluates all available genes and sifts only one gene at a time with the computation of minimum error rate. A gene is selected into the feature subset if it combines with the selected genes in a feature vector and these combined features can gain better classification information. Based upon the method, the classifier is constructed accordingly. Compared with some other machine learning methods, the experimental results show that the classifiers induced based on the proposed method are capable of generating accurate and interpretable results in biomedical applications.
Keywords
Bayes methods; cancer; genetics; medical computing; pattern classification; set theory; Bayes minimum error probability; SFBMEP; biomedical applications; cancer classification; classification information; classifiers; dynamic gene evaluation; feature selection; feature subset; feature vector; functional genes; gene selection; minimum error rate; Bayesian methods; Breast; Medical diagnostic imaging;
fLanguage
English
Publisher
ieee
Conference_Titel
Fuzzy Systems and Knowledge Discovery (FSKD), 2012 9th International Conference on
Conference_Location
Sichuan
Print_ISBN
978-1-4673-0025-4
Type
conf
DOI
10.1109/FSKD.2012.6233728
Filename
6233728
Link To Document