基于支持向量机的结肠癌信息基因提取  

The Extraction for Informative Genes of Colon Cancer Based on Support Vector Machine

作　　者：李烨[1] 王永丽[1] 贺国平[1] 

机构地区：[1]山东科技大学信息科学与工程学院,山东青岛266590

出　　处：《山东科技大学学报（自然科学版）》2012年第3期84-89,共6页Journal of Shandong University of Science and Technology(Natural Science)

基　　金：国家自然科学基金项目(10971122);山东省优秀中青年科学家科研奖励基金项目(2010BSE06047);高等学校博士学科点专项科研基金项目(20093718110005);山东省科技攻关计划项目(2009GG10001012);山东省自然科学基金项目(Y2008A01)

摘　　要：基于结肠癌基因表达谱数据集,提出了一种信息基因提取的新方法。该方法结合了支持向量机(SVM)、Bhattacharyya距离、递归特征消除(RFE)和快速基于相关性过滤器(FCBF)方法。首先,利用Bhattacharyya距离与SVM-RFE方法结合去除无关基因,然后运用FCBF方法得到信息基因,最后以支持向量机作为分类器对结肠癌样本进行分类识别。实验结果表明,同现有的方法相比,该方法在提取基因数量和准确率上都有明显的优势。A new informative gene extracting method was developed based on the dataset of expressible spectrum in colon cancer genes and the method combined with the support vector machine,Bhattacharyya distance,recursive feature elimination(RFE) and fast correlation-based filter(FCBF).Firstly,a combination of Bhattacharyya distance and SVM-RFE was applied to remove the irrelevant genes.Secondly,FCBF was employed to generate the informative genes.Finally,SVM was used to classify the colon datasets.The experimental results demonstrated that compared to existing methods,the classification accuracy was improved and the number of informative genes was decreased.

关 键 词：结肠癌 支持向量机 信息基因 BHATTACHARYYA距离 递归特征消除 快速基于相关性过滤器 

分 类 号：O224[理学—运筹学与控制论]