计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (28): 209-211.

• 图形、图像、模式识别 • 上一篇    下一篇

基于自适应最近邻的局部保持投影算法

喻 军,李志勇,孟金涛   

  1. 郑州航空工业管理学院 数理系,郑州 450015
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-10-01 发布日期:2011-10-01

Locality preserving projections based on adaptive nearest neighbor

YU Jun,LI Zhiyong,MENG Jintao   

  1. Department of Mathematics and Physics,Zhengzhou Institute of Aeronautical Industry Management,Zhengzhou 450015,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-10-01 Published:2011-10-01

摘要: 局部保持投影(LPP)是一种新的数据降维技术,但其本身是一种非监督学习算法,对于分类问题效果不是太好。基于自适应最近邻,结合LPP算法,提出了一种有监督的局部保持投影算法(ANNLPP)。该方法通过修改LPP算法中的权值矩阵,在降维的同时,增加了类别信息,是一种有监督学习算法。通过二维数据可视化和UMIST、ORL 人脸识别实验,表明该方法对于分类问题具有较好的降维效果。

关键词: 监督学习, 自适应最近邻, 局部保持投影

Abstract: Locality Preserving Projections algorithm(LPP) is a new dimensionality reduction technique.But it is an unsupervised learning algorithm.It could not process classification effectively.A new supervised learning algorithm is proposed,which combines with adaptive nearest neighbor and LPP.The method modifies the similar measure matrix of LPP.Compared with LPP,because it uses class information fully,the algorithm is a supervised learning.Experimentation based on 2-D visualization and UMIST,ORL face datasets shows that the proposed method achieves higher recognition rate.

Key words: supervised learning, adaptive nearest neighbor, locality preserving projections