基于类别保留投影的基因表达数据特征提取新方法  

New Method of Feature Extraction for Gene Expression Data Based on Class Preserving Projection

作　　者：王文俊[1] 

机构地区：[1]西安电子科技大学计算机学院,陕西西安710071

出　　处：《电子学报》2012年第2期358-364,共7页Acta Electronica Sinica

基　　金：中央高校基本科研业务费专项资金(No.K5051203013)

摘　　要：从两两样本的类别关系出发,提出一种新的线性鉴别特征提取方法,叫做类别保留投影.相比经典的fisher线性鉴别分析方法,类别保留投影具有最优子空间维数不受样本类别数限制、计算复杂度低的优点.通过对真实基因表达数据进行样本分类识别,证实了本文方法的有效性.并将类别保留投影方法推广到非线性空间,提出核类别保留投影,用于解决非线性特征提取问题,对基因表达数据的实验验证了方法的可行性.A new method of discriminant feature extraction, called Class Preserving Projection （CPP）, is proposed from the point view of class relation of pairwise samples. Compared to LDA, CPP has the following two advantages. One is that the optimal subspace dimension is not restricted to the number of categories of samples, and the other is that computational complexity is lower. Experiments are performed on gene expression data for sample classification, and the results confirm the effectiveness of the method. Kernel CPP （KCPP） is presented by generalizing CPP to nonlinear space to solve the problem of nonlinear feature extraction, and the experiments on gene expression data verify the feasibility of the method.

关 键 词：特征提取 FISHER线性鉴别分析 小样本 基因表达数据 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]