Presentation | 1995/9/28 A Study on Feature Selection for Small Class Classification Problems Tetsushi WAKABAYASHI, Shinji TSURUOKA, Fumitaka KIMURA, Yasuji MIYAKE, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Although the discriminant analysis has been most widely known as a typical statistical feature selection technique, its availability to small class classification problems is limited because the number of selectable features can not be or exceed the number of classes. In order to remove the limitation, a new feature selection technique ( F-K-L ) is proposed and tested by a handwritten numeral recognition experiment. While the discriminant analysis maximizes the variance ratio ( F-ratio ), and the principal component analysis ( K-L expansion ) minimizes the square error of representing a mixture distribution of total class, the F-K-L optimizes both the F-ratio and the square error simultaneously. The result of experiment shows that the F-K-L provides the richest features in discriminating power for the small class classification problems when compared with other techniques including the discriminant analysis, the principal component analysis, and the orthonormal discriminant vector method ( ODV ). |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | feature selection / feature extraction / canonical discriminant analysis / principal component analysis / K-L expansion |
Paper # | PRU95-115 |
Date of Issue |
Conference Information | |
Committee | PRU |
---|---|
Conference Date | 1995/9/28(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 | Pattern Recognition and Understanding (PRU) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study on Feature Selection for Small Class Classification Problems |
Sub Title (in English) | |
Keyword(1) | feature selection |
Keyword(2) | feature extraction |
Keyword(3) | canonical discriminant analysis |
Keyword(4) | principal component analysis |
Keyword(5) | K-L expansion |
1st Author's Name | Tetsushi WAKABAYASHI |
1st Author's Affiliation | Department of Information Engineering, Faculty of Engineering, Mie University() |
2nd Author's Name | Shinji TSURUOKA |
2nd Author's Affiliation | Department of Information Engineering, Faculty of Engineering, Mie University |
3rd Author's Name | Fumitaka KIMURA |
3rd Author's Affiliation | Department of Information Engineering, Faculty of Engineering, Mie University |
4th Author's Name | Yasuji MIYAKE |
4th Author's Affiliation | Department of Information Engineering, Faculty of Engineering, Mie University |
Date | 1995/9/28 |
Paper # | PRU95-115 |
Volume (vol) | vol.95 |
Number (no) | 278 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |