摘要
在构造仿射矩阵时,满足稀疏性就会降低其分组效应,反之,又不利于数据的选择.针对这些问题,提出投影相关自适应子空间分割方法.通过引入迹lasso,自适应地根据样本数据的相关性构造仿射矩阵,同时提取出有利于类别识别的特征.在6个基因表达数据上的实验结果表明,该方法优于现有子空间分割方法.
As constructing the affine matrix, satisfying data sparsity will reduce its group effect, on the contrary, it does not help to select data. To solve these problems, we propose projection correlation adaptive subspace segmentation （PCASS）. Trace lasso can determine affine matrix adaptively by the correlation of the sample data, and can extract features those are useful to discriminate types as well. Experimental results on six gene expression data show that this method is superior to the other existing subspace segmentation.
作者
陈慧娟
陈晓云
CHEN Huijuan CHEN Xiaoyun(College of Mathematies and Computer Seienee, Fuzhou University, Fuzhou, Fujian 350116, China)
出处
《福州大学学报（自然科学版）》
CAS
北大核心
2017年第1期44-49,共6页
Journal of Fuzhou University(Natural Science Edition)
基金
福建省自然科学基金资助项目(2014J01009)
关键词
子空间分割
基因表达数据
降维
subspace segmentation
gene expression data
dimension reduction
