A Model of Solving of Pattern Recognition Problems Based on Resolution Method

Abstract

Deductive and inductive approaches to solving of pattern recognition problems are considered. Logical and precedent-related encoding models of prior information are used in these approaches correspondingly. Algebra of objects is built and its isomorphism to Boolean algebra is shown. Algorithms for mutual conversion of encoding models are developed. A modification of the resolution method for solution of pattern recognition problems is suggested. An algorithm combining the resolution method and a parametric family of recognition algorithms is developed.