@article{oai:bunkyo.repo.nii.ac.jp:00003353,
 author = {鈴木, 昇一},
 journal = {情報研究, Information and Communication Studies},
 month = {1989-01-01, 2012-01-17},
 note = {In a mathematical theory of recognizing patterns which has been presented by S. Suzuki it is fundamentally important to construct in an adaptable manner a contraction-mapping, a similarity-measure function and a rough classifier which must satisfy axiom 1, 7 and 9 respectively so that a system may recognize patterns by the use of a structural fertilization of fixed-point type as correctly as possible. We shall explain a method for finding a weight vector in the rough classifier which separates two given finite sets of infinite-dimensional patterns by means of a multivariate analysis and illustrate the results with simulations of the method which demonstrates a separability of the two set of the rough classifier. It is concluded that the computer implementation of the multivariate analysis is satisfactory to some degree in the case of linear feature-extracting elements.},
 pages = {35--49},
 title = {多変量解析に基づく大分類関数の決定とその計算機シミュレーション},
 volume = {10},
 year = {}
}