Computer Engineering and Applications ›› 2011, Vol. 47 ›› Issue (36): 197-202.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Improved local preserving projection algorithm based on exponential diagonal matrix

AN Yajing1,WANG Shitong2   

  1. 1.School of Internet of Things Engineering,Jiangnan University,Wuxi,Jiangsu 214122,China
    2.School of Digital Media,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-12-21 Published:2011-12-21



  1. 1.江南大学 物联网工程学院,江苏 无锡 214122
    2.江南大学 数字媒体学院,江苏 无锡 214122

Abstract: Local Preserving Projection algorithm(LPP) obtains the diagonal matrix by computing the sum of the Euclidean distance and it can make use of the results to reduce dimensions.However,it is worth to do further investigation whether the algorithm can be optimized.In the paper,some modification are made to the formula which the algorithm rests on.With adding the exponential parameter,the improved local preserving projection algorithm based on exponential diagonal matrix can be got.Through a lot of experiments it has an affect on the results of the dimensionality reduction.It is easier to approach the intrinsic dimension of the datas.It also checks the discrimination after reducing the dimensions and the sensitivity of noise.

Key words: dimensionality reduction, local preserving projection, diagonal matrix, exponential parameter, noise

摘要: 局部保持映射(LPP)算法利用欧几里德距离求得权值累加得到对角矩阵,利用结果进行降维。对于这个算法是否可以进一步优化还值得进一步探讨。对该算法所依据的公式进行修改,在对角矩阵上引入指数参数,形成对角距阵指数优化的局部保持映射算法。通过实验可以证明,对角距阵指数优化的局部保持映射算法能够影响降维的结果,可以使得降维更容易得到接近本征维数的投影向量,通过实验验证降维后的识别效果和对噪声的敏感度。

关键词: 维数约简, 局部保持映射(LPP), 对角矩阵, 指数参数, 噪声