Search

The purpose of this paper is to review models of autoassociative, distributed memory that can be naturally implemented by neural networks (mainly with local learning rules and iterative dynamics based on information locally available to neurons). В настоящем обзоре рассмотрены модели автоассоциативной распределенной памяти, которые могут быть естественным образом реализованы нейронными сетями. У цьому огляді розглянуто моделі автоасоціативної розподіленої пам’яті, які можуть бути природним чином реалізовані нейронними мережами.

The cybernetics approach is extended to systems of biological nature. The contribution of biological cybernetics and medical cybernetics to the extension of capabilities of solving theoretical and applied problems of biomedicine is demonstrated by concrete examples. Telemedicine and virtual concilia are considered in the structure of a unified medical information space.

The solution of a system of problems is modeled as a search among alternative paths on a semantic network. We consider representation of procedural knowledge and modeling of decision processes by production rules and Petri nets.

Our method enables the responses of primary converters to be linearized in a practical way in the context of optimal filtering of input signals that reflect the measured parameters in the vast majority of technical objects. The circuits provide faithful and reliable restoration of measured experimental data without extra consumption of working memory and computer time on routine conversion of calibration characteristics. This is extremely important when developing digital computing complexes and real time systems, using large digital computers and minicomputers.