@article{oai:bunkyo.repo.nii.ac.jp:00003323,
 author = {鈴木, 昇一},
 journal = {情報研究, Information and Communication Studies},
 month = {1985-01-01, 2012-01-17},
 note = {Many people believe that a mathematica l theory does not exist even now which can treat patterns to be recognized using a unified approach in the whole field of patternrecognition, i.e.preprocessing, feature-extraction, classification, etc..In this paper it is examined some examples of contraction mapping T : Φ→Φ which has started S.Suzuki building up the mathematical theory of recognizing patterns.
　Mapping T has four properties. The most important property of them is an idempotent law T ? T =T.Five examples of T are presented : (1)reduced structural-model mapping (2)sampling (3)projector (4)band- limited (5)quantization (6)operator preserving an average similarity measure. By giving meanings of those examples we shall make it clear how much mapping T plays an fundamentally important role on the scene of obtaining a reduced representation of an input pattern to be recognized having various information.},
 pages = {19--30},
 title = {収縮写像に関する一考察},
 volume = {6},
 year = {}
}