Abstract: In this paper the calculating estimates model, based on a system of logical regularities to solve the supervised classification problem is considered. A method of determining the weighting parameters of the decision rule, as well as two methods to eliminate bounce on the classification by introducing a logical disjunction of the negations of regularities and the approximation of regularities by sigmoid logic functions are proposed. The results of numerical experiments are presented.

Keywords: classification, pattern recognition, estimates calculation, multi-algorithmic approach, logical regularities

ACM Classification Keywords: I.2.4 Artificial Intelligence Knowledge Representation Formalisms and Methods – Predicate logic; I.5.1 Pattern Recognition Models – Deterministic, H.2.8 Database Applications, Data mining; I.5.3 Clustering.

Link:

ОБ ОДНОМ МЕТОДЕ МУЛЬТИАЛГОРИТМИЧЕСКОЙ КЛАССИФИКАЦИИ

(On a method of multi-algorithmic classification)

Сергей Львов, Владимир Рязанов

http://www.foibg.com/ijita/vol22/ijita22-02-p03.pdf