Vol 11, No 5 (2010)

В сентябре 2010 года Красноярск принимал участников XII Международного симпозиума по непараметрическим методам в кибернетике и системному анализу. Его организаторами были Сибирский государственный аэрокосмический университет имени академика М. Ф. Решетнева, Национальный исследовательский Томский государственный университет, Институт вычислительного моделирования СО РАН. В настоящий «Вестник» вошли некоторые работы, представленные на симпозиуме. Симпозиум был посвящен развитию современных математических методов для создания интеллектуальных компьютерных систем различного назначения, функционирующих в условиях неполных данных об исследуемом процессе. Обсуждались также проблемы системного анализа. Тематика симпозиума представлена следующими секциями: • ·непараметрическая и робастная статистика; • ·адаптивные и обучающиеся системы; • ·непараметрические системы управления; • ·оптимизация и принятие решений; • идентификация и моделирование; • ·распознавание образов, классификация и обработка изображений; • ·проблемы системного анализа; • ·управление экономическими и организационными системами; • ·искусственный интеллект и экспертные системы; • ·разработка и применение компьютерных систем различного назначения и систем автоматизации. Красноярск не впервые становится местом проведения симпозиума; проходил он также и в Томске, Иркутске, Железногорске, Дивногорске, Шушенском. Традиционно в первой половине дня заслушивались два пленарных доклада, во второй половине дня - секционные доклады. Результаты научных исследований широко обсуждались участниками симпозиума в течение всего времени его проведения. Во время работы симпозиума спонтанно возникали дискуссии по различным проблемам кибернетики, информатики, системного анализа, менеджмента, образования. Материалы одной из дискуссий, посвященной образовательному процессу в нашей стране, приведены в данном «Вестнике». Считаем необходимым отметить большую и эффективную работу доцентов кафедры системного анализа и исследования операций СибГАУ Е. Д. Агафонова и О. В. Шестер-невой, аспиранта кафедры А. С. Орловой, студентов кафедры Л. Н. Голуб и Д. В. Безмен по организации и проведению симпозиума и подготовке материалов к опубликованию. Профессор А. В. Медведев Профессор Ф. П. Тарасенко

In this article the author considers the term «culture» which has different content in different areas of human activity and gives analysis to interconnection of models, modeling and culture of an individual. Some aspects of immersion of individuals in culture, models and environment interaction are given as well.

Problems of statistical forecasting are considered for autoregressive time series observed under interval censoring. Optimal forecasting statistic is proposed, its mean-square risk is evaluated. Comparison of optimal and widely used in practice forecasting statistics is made. Numerical results are given.

In this paper we describe and investigate a method which allows us to detect the most informative features out of all data extracted from a certain data corpus. The significance of input features is estimated as an average absolute value of the first-order partial derivative. The method requires the values of the objective function at the certain assigned points. If there is no possibility to calculate these values (the object is not available for experiments), we use non-parametric kernel regression to approximate them. The algorithm is tested on different simulated objects and is used for investigation of the dependency between linguistic features of spoken utterances and speakers ' capabilities.

Tn the article we propose algorithms for search of the set quantity of the principal minimum of multiextremal functions of continuous variables. The basis of the algorithms is sequential search of global extremum based on coordinates averaging method, with subsequent isolation of the found subdomains with the help of additional inequalities constraints. On numerical examples we demonstrate the overall effectiveness of the algorithms.

In the article we consider an algorithm of nonparametric generator's building for stohastic stationary process. Dependency interval of stochastic process is determined with the help of nonparametric algorithms ofprognosis.

The paper describes application of orthogonal series method for construction of controlled systems models under non-parametric uncertainty. A key element of the method is draw of orthogonal expansion length based on observations, in other words, defining parametric structure of the model. The method is demonstrated for estimation of distribution density and regression function. Directions for generalizing onto multi-dimensional case are also presented.

Features and options affecting the activities of organizations, ongoing formulation of the problem of organizational system modeling and managing are considered. Some non-parametric mathematical models of organizational systems are given.

In the article we consider nonparametric methods of estimation of a regression and its derivatives on samplings of random variables with some singularities at their measurement. A bootstrap-method applied to the decision of the passes filling task in incomplete data or elimination of emptiness in space of observations is presented.

The algorithm of random value selection with adjusted distributions generation and numerical characteristics is considered (mathematical estimation, dispersion etc.).The algorithm proceedings of high precisions is showed on calculated examples of selection modeling. The selection generation common concept allows to extract this approach to the special category of random value generation algorithms. It was joined to the term "The Random selection precision generator".

A problem of image reconstruction is considered. Nonlinear Gaussian filter, built on the basis of the spectral matrix norm, is described. Results of numerical experiments confirming advantage of the given filter in comparison with linear Gaussian filter are discussed.

The results of computer studies of time series composed of the Wolfer's numbers are presented in the article. In accordance with the approach based on Takens's theorem, the author solves the problem of reconstructing dimension of attractor of a dynamical system describing the dynamics of the Sun. It is shown that for the last 3 000 years, fundamental changes in the evolution of the Sun have not happened.

In this paper the analysis of efficiency of automatically adjusting genetic algorithm and adaptive hybrid stochastic procedure has been fulfilled on representative set of test and real world economical problems. It has been proved that the developed modifications outperform the standard genetic algorithm.

In the article a method of estimation of influence of additive disturbance with various laws of distribution on an input of the receiver of telecommunication system according to combined parameter of identification for signals with modulation by the minimal frequency shift is considered.

We consider a computer simulation of hydrothermal and ice regimes of the Angara river up and downsream from the dam of the Boguchanskaya HPP construction.

A mathematical model for the processes of mult'component mixtures separation 's presented. A рroblет of optimal control is formulated. Necessary optimal conditions are obtained. A numerical method of optimization problems solving is developed.

In this article the author presents a two-dimensional model of filtering, taking into account movement of air and hydrocarbon pollutants in porous soil. The model includes a system of partial differential equations with additional conditions. The differential equations include mass balance equation element in a porous medium - inseparability equation, and the differential equations of motion. For circuit system the author puts in the equation of state of pollutant and environment. Initial and boundary conditions correspond to the filtration process, beginning with the ground surface and initial stage of oil products spillage. A comparative analysis of the results of mathematical modeling with experiments is presented.

Typical modes of hardware-software means complexing in complex systems of protection of the information are allocated, the mathematical basis for definition of their dynamic characteristics is formed.

In the article we consider experience and prospects of application of the system approach to management of human resources. Results of experiment on system design of system of training in organization are elaborated with offers on application of other approaches connected with features of a person as the biological and social subject. Tools of consideration are the synergetics, reflexive theatre and Tai-Chi as an example of approach based on oriental philosophy.

Task of regression function restoration according observations with random errors is considered in the article. Usage of the classical nonparametric estimations of regression curve for such purposes doesn't always give satisfactory results on small samples. The new class of nonparametric estimations is proposed, it allows to increase the regression curve restoration quality of according observations.

The paper considers a problem of control of linear dynamic system in conditions of nonparametric uncertainty. A type of equation describing the process, is indeterminate. Estimation of reverse operator of the dynamic process as the algorithm of control is offered. A nonparametric model of a dynamic system is realized on the basis of its transfer features. The results of the research of the nonparametric algorithm of control by a linear dynamic system using the method of statistic modeling.

In the paper the authors consider a task of creation of a control object transfer function in accordance with a mathematical model of closed linear dynamic system for a regulator parameters tuning. Structure synthesis of a closed loop model is carried out with nonparametric modeling methods.

In paper it is presented concept of scene image construction by way of video sequence frames or photo shot series superposition. Frame superposition based on construction of disparity map imaged vectors of shift blocks

The task offrequency test of sandwich plate wrinkling vibrations, when plate is composed of two similar composite bearing layers and orthotropic filler, is solved in this paper. Hamilton's principle was applied to derive a basic differential equation of forth order. Frequency formula for symmetrical vibrations of sandwich plate with pinned bearing layers is given.

Most important characteristics of cryptographic functions are balancedness, nonlinearity, propagation criterion, correlation immunity, degree and non-existence of nonzero linear structure. Partially-bent functions form super-class of the class of bent functions. These /unctions may achieve desirable characteristics. Two algorithms for generation of partially-bent functions were supposed and studied. Second algorithm may improve cryptographic characteristics of generated functions.

In geometrically nonlinear arrangement a variational problem of deformation of shell with low value of transverse shear stiffness is considered. The conclusion of expression of potential energy of deformation which can be used for generation of the equations of stability for shells with low value of transverse shear stiffness.

It is represented new method of passive refractive optical SD scanning which is based on preliminary preprocessing of measured objects by creation on its surfaces of artificial textures and post processing on a different level of details.

Application of logarithmic amplifier to design the microwave range complex reflection coefficient (CRC) measuring device of terminals is described in this article. The CRC measuring methods using the sensor, which has the parameters precising during the measuring, are considered there. The measurement error sources are also analyzed.