基于光滑近邻表示的基因表达数据子空间聚类

Gene expression data subspace clustering based on smooth neighbor representation

基因表达数据具有样本数少、基因维数高、非线性等特点,为能有效地处理基因表达数据,提出光滑近邻表示子空间聚类算法.利用每个数据点的近邻线性表示刻画数据集的非线性特点,并对近邻表示添加光滑约束,使数据点与近邻的距离关系嵌入到该数据点的重构表示中.在基因表达数据上的实验表明,所提出的方法优于其他几个现有方法,进而表明所提出方法对基因表达数据的聚类是有效的.

Gene expression data has the characteristics of small sample size, high dimension, nonlinear and so on. In order to effectively deal with the gene expression data,a subspace clustering method is proposed via smooth neighbour representation（SNR）. The neighborhood linear representation of data points is used to describe the nonlinear properties of data, and the smooth constraint is added on the representation which makes the relationship of distance between data point and its neighbors embed in the reconstruction representation. Experiment results on gene expression data show that the performance of SNR is superior to several existing methods, and SNR can cluster gene expression data effectively.