计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (28): 181-183.

• 数据库与信息处理 • 上一篇    下一篇

一种在源数据稀疏情况下的数据降维算法

宋 欣1,3,叶世伟2   

  1. 1.中国科学院 研究生院 工程教育学院,北京 100049
    2.中国科学院 研究生院 信息科学与工程学院,北京 100049
    3.东北大学 秦皇岛分校,河北 秦皇岛 066004
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-10-01 发布日期:2007-10-01
  • 通讯作者: 宋 欣

Data dimensionality reduction algorithm when source data is spare

SONG Xin1,3,YE Shi-wei2   

  1. 1.College of Engineering of the Graduate School of the Chinese Academy of Sciences,Beijing 100049,China
    2.School of Information Science and Engineering of the Graduate School of the Chinese Academy of Sciences,Beijing 100049,China
    3.Northeastern University at Qinhuangdao,Qinhuangdao,Heibei 066004,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-10-01 Published:2007-10-01
  • Contact: SONG Xin

摘要: 流形学习方法是根据流形的定义提出的一种非线性数据降维方法,主要思想是发现嵌入在高维数据空间的低维光滑流形。从分析基于流形学习理论的局部线性嵌入算法入手,针对传统的局部线性嵌入算法在源数据稀疏时会失效的缺点,提出了基于局部线性逼近思想的流形学习算法,并在S-曲线上采样测试取得良好降维效果。

关键词: 降维算法, 局部线性逼近, 流形学习, 局部线性嵌入

Abstract: Manifold learning is a nonlinear data dimensionality reduction method.It is proposed according to the manifold concept.The main idea of manifold leaning is to find a smooth low-dimensional manifold embedded in the high-dimensional data space.The Locally Linear Embedding(LLE) algorithm based on Manifold learning is introduced firstly in this paper,because the
conventional LLE algorithm will be ineffective when the source data is spare.,with that the manifold learning algorithm based on Locally Linear Approximating(LLA) is presented.At last,the results show the effectiveness of the LLA on the S-curse sampling
and testing.

Key words: dimensionality reduction algorithm, Locally Linear Approximating(LLA), manifold learning, Locally Linear Embedding(LLE)