Selecting Informative Variables in the Identification Problem
Скачать файл:
URI (для ссылок/цитирований):
http://journal.sfu-kras.ru/number/29983https://elib.sfu-kras.ru/handle/2311/69533
Автор:
Eugene D. Mihov
Oleg V. Nepomnyashchiy
Коллективный автор:
Институт космических и информационных технологий
Кафедра вычислительной техники
Дата:
2016-12Журнал:
Journal of Siberian Federal University. Mathematics & PhysicsКвартиль журнала в Scopus:
Q4Библиографическое описание:
Eugene D. Mihov. Selecting Informative Variables in the Identification Problem [Текст] / Eugene D. Mihov, Oleg V. Nepomnyashchiy // Journal of Siberian Federal University. Mathematics & Physics. — 2016. — Т. 9 (№ 4). — С. 478-485Аннотация:
The problem of multidimensional object classification with small training sample is considered. The
following algorithms of estimating variable informativeness are considered: Ad, Del, AdDel.
A new algorithm for selecting informative variables is proposed. It is based on the optimization of
the coefficient vector of the kernel fuzziness. Some modification of this algorithm is also discussed.
The comparative analysis of existing methods for selecting informative variables is presented.