| 학회 | 한국화학공학회 |
| 학술대회 | 2004년 봄 (04/23 ~ 04/24, 공주대학교) |
| 권호 | 10권 1호, p.170 |
| 발표분야 | 공정시스템 |
| 제목 | Comparative studies of marker genes selection criteria for cancer classification |
| 초록 | For the correct diagnosis, selection of marker genes with good discrimination power is crucial before making a classifier. Various selection methods have been proposed, and they have been tested in classification accuracy, computational time, and robustness of noise. On that ground, it is generally believed that there is no universal criterion that is superior over the rest. Nevertheless, many researchers are trying to find a powerful selection criterion which can be generally applied In order to save time and cost for a good classifier. In this work, we compared recent criteria for marker genes selection. New metric, SVMs/GAs, and MAximal MArgin Linear Programming (MAMA) method were examined. They are different from conventional methods in its aspect of selecting mechanism. They are not based on differences of gene expression level. As a results, they are superioir to conventional methods. |
| 저자 | 이민영1, 황대희2, 이인범3, 한종훈1 |
| 소속 | 1포항공과대, 2The Institute for Systems Biology, 3USA |
| 키워드 | microarray; cancer diagnosis; marker gene |
| 원문파일 | 초록 보기 |
