基于双正交非负矩阵三因式的肿瘤识别  

Tumor recognition with bi-orthogonal nonnegative matrix tri-factorization

作　　者：谭青青[1] 王年[1] 苏亮亮[1] 方正文[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039

出　　处：《安徽大学学报（自然科学版）》2015年第6期29-36,共8页Journal of Anhui University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61172127);安徽省自然科学基金资助项目(1208085MF93);安徽大学211创新团队项目(KJTD007A;KJTD001B)

摘　　要：基因表达谱数据分析已经逐渐成为疾病诊断和分类的常规步骤.目前人们对NMF(nonnegative matrix factorization)的大多数研究都专注于二因式分解.论文另辟蹊径,对BONMTF(bi-orthogonal nonnegative matrix tri-factorization)算法进行了系统化的分析,利用此算法得到表征样本属性的矩阵,并将其应用于基因表达谱数据分析,提高了样本识别率.实验采用4组具有代表性的肿瘤基因表达谱数据,其结果证明了论文方法针对不同数据集的识别率都比传统方法有所提高,具有一定的可行性及应用前景.Analysis of gene expression profile data has gradually become a routine procedure for disease diagnosis and classification. Currently, most research on NMF focused on two- factor factorization. This paper found a new path. It presented systematic analysis of BONMTF, then it used the algorithm to get the matrix which characterizes properties of sample. Furthermore, this paper also applied BONMTF algorithm in the analysis of gene expression profile data. Therefore, it improved the recognition rate of samples. Four representative groups of gene expression data were used for test. The result proved that, for different data sets, the proposed method had higher recognition rate than conventional methods. Therefore,it had certain feasibility and application prospects.

关 键 词：三因式分解 双正交非负矩阵 肿瘤识别 基因表达谱数据 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]