Kernel MT system and its application  

Oleh:

Sano, Natsuki ; Kobayashi, Yuta ; Kuroki, Manabu ; Etoh, Minoru

Jenis: Article from Proceeding
Dalam koleksi: ANQ Congress 19-22 October 2010 New Delhi, page 1-10.
Topik: Mahalanobis-taguchi system ; Mahalanobis distance ; Kernel function
Fulltext: JSQC_Natsuki_Sano_Kernel_MT_system_and_Its_application.pdf (172.89KB)

Isi artikel

The Mahalanobis-Taguchi System (MTS) is one of statistical methods for analyzing asymmetric discriminant data in the field of quality control and medical diagnosis. It is pointed out, however, that the accuracy of the MTS is deteriorated when (i) the data generating process of a given normal group is not based on a Gaussian distribution, and (ii) some of variables are highly correlated with each other. In order to solve those problems, we propose a new asymmetric discriminant analysis, the kernel MTS, which is formulated by introducing a kernel technique used in the field of machine learning into the Type I MT distance proposed by Miyakawa and Nagata. Taking a discrimination problem between premium wine and normal wine as an experiment, the experiment shows that the kernel MTS is superior to the MTS in terms of odds ratios.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali