Eliciting and the Use of Information Concerning Regular Structures in the Formalism of Functional Neural Networks in Decision-Support Systems

Abstract

Methods of increasing the efficiency of systems working with knowledge bases and built in the formalism of functional neural networks are described. The occurrence of regular structures in the FN network in the context of substitutions from the knowledge base during the correction of the FN network in the search for a solution is considered. An economical way of presenting them is proposed, taking into account possible heterogeneity outside the context of substitution, as well as methods for reducing the volume of expansions of the FN network when searching for a solution using information about the identified regular structures. An algorithm is proposed that identifies permutations and allows, without costly manipulations with the knowledge base, to identify useful substitutions, operating with information about the positions of elements in a regular structure.