New computer components for proximity assessment and recognition of double objects, which are encoded with symbols of the binary alphabet
Abstract
Numerous binary measures of similarity and distance make it possible to effectively solve various tasks of recognition, classification, estimation of proximity of binary sequences, etc. Since the performance of the proposed systems largely depends on the choice of appropriate measures and distances, many researchers have spent considerable effort to find more efficient binary ratios for solving the mentioned tasks. Numerous binary relations, especially binary ones, have been proposed and investigated in various fields of science. However, it was not possible to determine one or a group of distances or similar measures effective in solving any tasks. In this regard, a number of new methods were proposed, based on the use of algorithms for learning, and on the external similarity of ratios that have different names of methods and proposed by different specialists, and the introduction of pre-processing of the initial data in order to determine the weighting factors when using them. Il.: 2; Bibliogr.: 15 titles.