Presentation 2007-02-22
Semi-Supervised Locality Preserving Projection with Clustering Mapping
Weiwei DU, Kiichi URAHAMA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We present a semi-supervised classification method where data are mapped into a classification subspace and classified with the nearest neighbor rule. Training data are preliminarily fuzzy-clustered and prototype data of clusters are labeled. Data are then mapped in two stages. In the first stage, data are mapped into membership vectors from which multiplicatively weighted similarities between data are evaluated. In the second stage, the membership vectors are linearly projected into a classification subspace by using the locality preserving projection with the weighted similarities. The dimensionality of the subspace is the number of classes minus 1. Test data are similarly projected with the 2-stage mapping and classified to the class of nearest labeled training datum. We examined the classification rates of the projection method in the second stage and then show that the 2-stage mapping method can classify data of arbitrarily-shaped clusters such as face images.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) semi-supervised learning / fuzzy clustering / locality preserving projection / RBF mapping
Paper # PRMU2006-211,HIP2006-104
Date of Issue

Conference Information
Committee PRMU
Conference Date 2007/2/15(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Semi-Supervised Locality Preserving Projection with Clustering Mapping
Sub Title (in English)
Keyword(1) semi-supervised learning
Keyword(2) fuzzy clustering
Keyword(3) locality preserving projection
Keyword(4) RBF mapping
1st Author's Name Weiwei DU
1st Author's Affiliation Faculty of Design, Kyushu University()
2nd Author's Name Kiichi URAHAMA
2nd Author's Affiliation Faculty of Design, Kyushu University
Date 2007-02-22
Paper # PRMU2006-211,HIP2006-104
Volume (vol) vol.106
Number (no) 538
Page pp.pp.-
#Pages 6
Date of Issue