COMPLEX-FUNCTIONAL BLOCKS OF COMPUTATIONAL PERCEPTION IN THE CLASS OF COMPUTATIONAL-TRANSFORMATIVE CIRCUITS
Based on the set of generalized properties of the physical character system (FSS), the structural-analog architecture of computational perception is considered. Different levels of abstraction of a natural-like object are involved. The use of archetypes in the classification of computing-conversion circuits (VP-circuits) on operating with controlled parameter elements provides the possibility of systematization of not only available, but also conceivable structures. Based on the ontological interpretation of the structural and functional model, its quality is determined by the quality of the concepts and the relationships between them. In the conditions of destabilization of the internal and external environment, ensuring the equilibrium state of the target is provided by the mechanism of homeostasis. For the mathematical description of the structure of VP-circuits as stream devices of neural network architecture the results of approximation of reproducible dependences «input–output» in the class of rational functions (by the parameter of control of conductivity of operating elements) are used. When describing a class of objects and sets of mappings (morphisms) that translate one object into another, the theory of categories is involved. The General problem of synthesis is reduced to generation of a complete set of non-isomorphic cyclically connected irreducible structures of s-models of VP-chains. The procedure of synthesis of the minimum rank of the s-model of the VP-chain is given. The examples of original developments where the properties of computational perception in natural-like objects of artificial origin are used are discussed.
Authors: I. V. Gerasimov, S. A. Kuzmin, N. M. Safyannikov, A. V. Rudinsky
Direction: Informatics, Computer Technologies And Control
Keywords: Computational perception, computational-transform circuits, structural-analog computations, s-modeling, category theory, functor, structural number, relief virtualization, sensor, tremor
View full article