ISSN 2071-8594

Российская академия наук

 

abstracts

KNOWLEDGE ENGINEERING

V.L. Stefanuk. Creative solution of problems

An adequate problem representation is fundamental for its creative solution. For a systematic search for appropriate representation we introduced Semiotic Introspection Principle (SIP). It is illustrated with the examples of problems related to the “Tough Nut Problem” of AI.  In a more complex 3D version of the problem the heuristics SIP leads to a non-obvious (creative) solution literally in a few steps. It is assumed that the heuristics SIP is an important supplement to the well known heuristics GPS as SIP allows to consider problems, which in GPS require an extensive search.

Keywords: Creative Problem Solving, Problem Representation, Semiotic Introspection Principle, “Tough Nut Problem”, 3D variation.

A.G. Zhikharev, S.I. Matorin. Method for formalization of organizational knowledge

The paper considers a new method for representation of knowledge based on the original system graphical-analytical approach “Unit-Function-Object” (UFO-approach), the theory of patterns and calculation of processes. We describe a way for formalization of knowledge on the organization activity, which is represented by using the UFO-approach. Examples of application of the offered method are resulted.

Keywords: formalized representation of knowledge, organizational knowledge, systematic approach “Unit-Function-Object”, calculus of process, calculus of functions.

INTELLIGENT TECHNOLOGIES

V.I. Gorodetsky. A survey on Teamwork of Autonomous Agents: Theories, Frameworks and Specification Languages (Part 1)

Autonomous agents and multi–agent systems research are stimulated, in many respects, by modern robotics. Within the latter, one of the hottest problems is coordination of teamwork of autonomous agents intended for execution of a mission. At the same time, this aspect of multi–agent systems is the source of a number of the most difficult tasks. The paper is devoted to a brief survey on and analysis of the current state–of–the–art concerning with theories (part 1), frameworks and supporting reusable software, as well as to a comparative overview of the languages aimed at specification of the teamwork aspects of autonomous agents (part 2).

Keywords: autonomous agent, teamwork, theory of teamwork, teamwork model, software tool, specification language.

S.P. Dudarov, P.L. Papaev, A.N. Kudryashov, Yu.A. Karibova. Cellular and Neural Network Models for Ecological Safety Problems

There is represented a general statement of cellular and neural network modeling in the paper. The basic concepts and principles of cellular and neural network models are considered, their classification is suggested. A scheme of simulation of medium state changes in the basic cell at a new time by known current state of it and of the cells of environment is given. The example of cellular and neural network modeling of atmosphere pollution in the result of accident at industrial object is shown.

Keywords: cellular and neural network model, cellular model, artificial neural network, modeling, free air pollution, accidental emission.

INTELLIGENT DATA ANALYSIS

S.Yu. Bolotova, S.D. Makhortov. Relevant backward inference algorithms based on solving the production logical equations

Production system backward inference algorithms based on equation solving in LP structures, a special class of algebraic systems, are presented. The strategy of a relevant backward inference is aimed at optimization of queries to external information source (either to a database or an interactive user). The concept of the equations is also applicable to knowledge verification. Advantages of new algorithms are demonstrated experimentally on the test knowledge bases.

Keywords: lattice, logical equation, production system, backward inference, knowledge verification.

Z.S. Bekyasheva, A.V. Pavlov, Statistical properties of the hypotheses, formed by neural network with interconnections, implemented by Fourier-holography technique

Dependence of statistical properties of hypotheses, formed by neural network with interconnections, implemented by Fourier-holography method, on the filtration due to the recording medium dynamical range limitation is considered. Numerical simulations results are presented.

Keywords: neural networks, Fourier-holography.