基于邻近线性表示的降维方法  

Dimensionality reduction method based on adjacent and linear representation

作　　者：高小明[1] 冯洋琴[1] 陈菲[1] 

机构地区：[1]西南科技大学计算机科学与技术学院,四川绵阳621010

出　　处：《计算机应用研究》2014年第8期2323-2325,2330,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61202044)

摘　　要：针对如何在高维数据集中寻找嵌入的低维流形结构问题,提出了基于邻近线性表示的降维方法,该算法克服了基于成对距离相似性降维算法和基于重建系数相似性降维算法的缺点。该算法具有封闭解,时间复杂度低。实验证明,与LE、ISOMAP和LLE算法相比,该算法抗噪能力强,对邻域大小选取不敏感,在采样数据比较少的情况下也能正常运行。In order to project the high-dimensional data points into their low-dimensional counterparts,this paper proposed an algorithm,namely dimensionality reduction method based on adjacent and linear representation which incorporated pairwisedistance based similarity with reconstruction coefficient based similarity for finding the latent low dimensional structure of data.The algorithm could work well even in under-sampled case and robust to the noise. The experimental results show that this algorithm performance well than LE,ISOMAP and LLE.

关 键 词：降维 流形学习 无监督 相似图 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]