|
OALib Journal期刊
ISSN: 2333-9721
费用：99美元
|
|
|
|
正交化近邻关系保持的降维及分类算法
DOI: 10.11834/jig.20090714
Keywords: 流形学习,近邻保持嵌入,线性近邻传递算法
Abstract:
针对近邻关系保持嵌入（NPE）算法易于受到降低后的维数影响，而且性能依赖于正确的维数估计的问题，提出了一种正交化的近邻关系保持的嵌入降维方法――ONPE。ONPE方法是使用数据点间的近邻关系来构造邻接图，假设每个数据点都能由其近邻点的线性组合表示，则可以通过提取数据点的局部几何信息，并在降维中保持提取的局部几何信息，迭代地计算正交基来得到数据的低维嵌入坐标。同时，在ONPE算法的基础上，利用局部几何信息，提出了一种在低维空间中使用标签传递（LNP）的分类算法――ONPC。其是假设高维空间中的局部近邻关系在降维后的空间中依然得到保持，并且数据点的类别可由近邻点的类别得到。在人工数据和人脸数据上的实验表明，该算法在减少维数依赖的同时，能有效提高NPE算法的分类性能。
| [1] | Tanenbaum J B,Silva V,Langford J C,A global geometric framework for nonlinear dimensionality reduction,SCIENCE,2000(12).
|
| [2] | Roweis S T,Saul L K,Nonlinear dimensionality rnduction by locally linear embedding,SCIENCE,2000(12).
|
| [3] | Belkin M,Niyogi P,Laplacian eigenmaps for dimensionality reduction and data representation,Neural Computation,2003(6).
|
| [4] | Yin J W,LiuXM,Feng Z L,A local tangent space alignment based transductive classification algorithm,Ulm,Germany; Springer,2006.
|
| [5] | Liu X M,Yin J W,Feng Z L,Incremental manifold learning via tangent space alignment,Ulm Germany ; Springer,2006.
|
| [6] | Zhang Z,Zha H,Principal manifolds and nonlinear dimensionality reduction via tangent space alignment,SIAM Journal ON SCIENTIFIC COMPUTING,2006(1).
|
| [7] | He X,Cai D,Yon S C,Neighborhood preserving embedding,北京,2005.
|
| [8] | He X,Niyogi P,Locality preserving projections,Cambridge,MA,USA:MIT press,2003.
|
| [9] | Cai D,He X,Orthogonal locality preserving indexing,Salvador,Brazil,2005.
|
| [10] | Cai D,He X,Han J,Orthogonal laplacianfaces for face recognition,IEEE Transactions on Image Processing,2006(11).
|
| [11] | Wang F,Zhang C,Label propagation through linear neighborhoods,Pittsburgh,Penn.,USA,2006.
|
| [12] | Wang F,Wang J,Zhang C,Semi-supervised classification using linear neighborhood prepagation,New York,USA,2006.
|
| [13] | Saul L K,Rowels S T,Think globally,fit locally:unsupervised learning of low dimensional manifolds,JOURNAL OF MACHINE LEARNING RESEARCH,2004(2).
|
| [14] | ORL & yale
|
Full-Text
|
|
Contact Us
service@oalib.com QQ:3279437679 
WhatsApp +8615387084133
|
|
