Presentation | 2012-06-15 A fast implementation of a subspace method using PCA-L1 Mariko HIROKAWA, Yoshimitsu KUROKI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | The subspace method generates subspaces for each class, and assigns an input data to the subspace that expresses most aptly. To generate the subspaces, you generally use PCA (principal component analysis), but it is sensitive to outliers. Owing to relieve the influence of outliers, one uses PCA-L1 (PCA based on Li-norm maximization) in exchange for computational loads. PCA-L1 needs initial vector for each basis, and to get the initial vector takes a lot of time. This paper proposes obtaining the initial vector using Gram-Schmidt orthogonalization. The proposed method reduces the computational loads on the subspace method using PCA-L1, and outperform the subspace method using PCA. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | subspace method / principal component analysis based on L1-norm maximization |
Paper # | SIS2012-12 |
Date of Issue |
Conference Information | |
Committee | SIS |
---|---|
Conference Date | 2012/6/7(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Smart Info-Media Systems (SIS) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A fast implementation of a subspace method using PCA-L1 |
Sub Title (in English) | |
Keyword(1) | subspace method |
Keyword(2) | principal component analysis based on L1-norm maximization |
1st Author's Name | Mariko HIROKAWA |
1st Author's Affiliation | Kurume National College of Technology() |
2nd Author's Name | Yoshimitsu KUROKI |
2nd Author's Affiliation | Kurume National College of Technology |
Date | 2012-06-15 |
Paper # | SIS2012-12 |
Volume (vol) | vol.112 |
Number (no) | 78 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |