计算机工程与设计
計算機工程與設計
계산궤공정여설계
COMPUTER ENGINEERING AND DESIGN
2010年
5期
1063-1065,1069
,共4页
二进制粒子群优化算法%集成分类器%基因表达谱%信息基因%肿瘤分类
二進製粒子群優化算法%集成分類器%基因錶達譜%信息基因%腫瘤分類
이진제입자군우화산법%집성분류기%기인표체보%신식기인%종류분류
binary particle swarm optimization%ensemble classifier%gene expression profile%informative gene%tumor classification
针对肿瘤基因表达谱样本少,维数高的特点,提出一种用于肿瘤信息基因提取和亚型识别的集成分类器算法.该算法根据基因的Fisher比率值建立候选子集,再采用相关系数和互信息两种度量方法,分别构造反映基因共表达行为和调控关系的特征子集.粒子群优化算法分别与SVM和KNN构成两个基分类器,从候选子集中提取信息基因并对肿瘤亚型进行分类,最后利用绝对多数投票方法对基分类器的结果进行整合.G.Gordon肺癌亚型识别的实验结果表明了该算法的可行性和有效性.
針對腫瘤基因錶達譜樣本少,維數高的特點,提齣一種用于腫瘤信息基因提取和亞型識彆的集成分類器算法.該算法根據基因的Fisher比率值建立候選子集,再採用相關繫數和互信息兩種度量方法,分彆構造反映基因共錶達行為和調控關繫的特徵子集.粒子群優化算法分彆與SVM和KNN構成兩箇基分類器,從候選子集中提取信息基因併對腫瘤亞型進行分類,最後利用絕對多數投票方法對基分類器的結果進行整閤.G.Gordon肺癌亞型識彆的實驗結果錶明瞭該算法的可行性和有效性.
침대종류기인표체보양본소,유수고적특점,제출일충용우종류신식기인제취화아형식별적집성분류기산법.해산법근거기인적Fisher비솔치건립후선자집,재채용상관계수화호신식량충도량방법,분별구조반영기인공표체행위화조공관계적특정자집.입자군우화산법분별여SVM화KNN구성량개기분류기,종후선자집중제취신식기인병대종류아형진행분류,최후이용절대다수투표방법대기분류기적결과진행정합.G.Gordon폐암아형식별적실험결과표명료해산법적가행성화유효성.
Due to the characteristic of small sample numbers and thousands of genes in tumor gene expression profile, an ensemble classifier algorithm is proposed. The candidate subset comprises of genes with higher Fisher ratio value and the feature subset that reflects the coexpression behavior of genes and regulation relationship is established by coefficients and mutual information. Particle swarm optimization, support vector machine and k-nearest neighbor method are combined to form two different base classifiers and their results are assembled by voting method. Experimental results acquired from the lung cancer subtype recognition confirm the feasibility and effectiveness of the proposed algorithm.
