“Informatics and Applications” scientific journal

Volume 5, Issue 1, 2011

Content   Abstract and Keywords   About Authors

TWO-PRIORITY SYSTEM WITH RESERVATION OF CHANNELS AND MARKOV INPUT FLOW.

• A. V. Pechinkin  IPI RAN, apechinkin@ipiran.ru

Abstract: The two-priority queueing system with trunk reservation and Markovian input flow of customers is considered. It is supposed that customers of each priority have two different phase-type distributions of their service times. Trunk reservationmeans that there are some channels which can be used only by customers of high priority. On the basis of researching the special general model, analytic relations that permit to calculate the main stationary characteristics of the queueing system were obtained.

Keywords:  queueing system; relative priority; reserving the channels

IMPROVEMENTS OF THE NONUNIFORM ESTIMATE FOR CONVERGENCE OF DISTRIBUTIONS OF POISSON RANDOMSUMS TO THE NORMAL DISTRIBUTION.

• S. V. Gavrilenko  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, gavrilenko.cmc@gmail.com

Abstract: The nonuniform estimates for convergence rate in the central limit theorem have been built. Using these structural improvements, it is shown that absolute constant in the nonuniformestimate for convergence rate in the central limit theorem for Poisson random sums is strictly less than similar constant in the nonuniform estimate for convergence rate in the classical central limit theorem and, assuming finite third moment, it does not exceed 22.7707. As a result, nonuniform estimates for convergence rate of the mixed Poisson, particularly, negative binomial, random sums have been built.

Keywords:  central limit theorem; convergence rate; nonuniform estimate; absolute constant; Poisson randomsum; mixed Poisson distribution

ON ACCURACY OF APPROXIMATIONS FOR STANDARDIZED CHI-SQUARED DISTRIBUTIONS BY EDGEWORTH–CHEBYSHEV EXPANSIONS.

• G. Cristoph  Institute for Mathematical Stochastics, Faculty of Mathematics, University of Magdeburg, Magdeburg, Germany, gerd.christoph@ovgu.de
• V. V. Ulyanov  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov Moscow State University, vulyan@gmail.com

Abstract: Astandardized chi-squared randomvariable Vn with n degrees of freedomis consideredwith Edgeworth–Chebyshev expansions, the computable error bounds of orders for approximations of Vn have been obtained. The results are useful in applications, in particular, in analysis of statistics ratio of probability features.

Keywords:  asymptotic expansions; error bounds; chi-squared distribution

STABILITY OF FINITE MIXTURES OF GENERALIZED GAMMA-DISTRIBUTIONS WITH RESPECT TO DISTURBANCE OF PARAMETERS.

• V. Yu. Korolev  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University; Institute of Informatics Problems, Russian Academy of Sciences, vkorolev@comtv.ru
• V.A. Krylov  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, vkrylov@cs.msu.ru
• V. Yu. Kuz’min  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, silencershade@gmail.com

Abstract: Using the example of the “contamination” model, in terms of uniform distance and Levy metric, estimates of the stability of finite mixtures of generalized gamma-distributions against small disturbance of parameters were obtained.

Keywords:  generalized gamma-distribution; finite mixture; Levy metric

ON THE ACCURACY OF THE NORMAL APPROXIMATION TO DISTRIBUTIONS OF POISSON RANDOM SUMS.

• Yu. S. Nefedova  Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, julia n@inbox.ru
• I.G. Shevtsova   Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, ishevtsova@cs.msu.su

Abstract: Two-sided bounds were constructed for the constant in the Berry.Esseen inequality for Poisson random sums of independent identically distributed random variables with finite absolute moments of order . The lower bounds were obtained for the first time. For the case , the upper bounds were sharpened, and the nonuniform estimates were proved.

Keywords:  central limit theorem; Poisson random sums; Berry.Esseen inequality; absolute constant; nonuniform estimate

INFORMATION TECHNOLOGY OF ACTIVE PARAMETRIC IDENTIFICATION OF STOCHASTIC QUASI-LINEAR DISCRETE SYSTEMS.

• V.M. Chubich  Department of Applied Mathematics, Faculty of Applied Mathematics and Computer Sciences, Novosibirsk State Technical University, chubich 62@ngs.ru

Abstract: Some theoretical and applied aspects of the active parametric identification of the Gaussian nonlinear discrete systems are considered for the first time. The original results are obtained for the case when the parameters of the mathematical models to be estimated appear in the state and observation equations; the initial conditions and covariance matrices of the dynamic noise and measurements errors were considered. An example of optimal parameter estimation of one model structure is shown.

Keywords:  parameter estimation; maximum likelihood method; optimal input signal design; Fisher information matrix; optimality criterion

AGENT MODELING OF TERRITORIAL SYSTEM DEVELOPMENT.

• K. S. Chirkunov  A. P. Ershov Institute of Informatics Systems, cyril.chirkunov@computer.org

Abstract: An agent system that reflects development of the country economy (building of new factories, raising of the overall level of income) based on the model of territorial system is considered. The elements of the model are presented in the form of agents, autonomous units capable to interact with each other.

Keywords:  agent negotiation algorithms; territorial system; simulation; economic zoning

TWO MODELS OF RESOURCE ALLOCATION UNDER THE ORGANIZATION OF INVESTMENT PROCESSES.

• P. V. Demin  State Educational Institution “Moscow Academy of the Labor Market and Information Technology,” pdemin@mail.ru

Abstract: Two examples of solving the problems that arise during the organization of the investment processes which are associated with the modernization of economy are considered. The first one relates with the problems while choosing for financing a project among the variety. The second one relates with the problem of distribution of the investment resources among the enterprises that are a part of the holding.

Keywords:  investment process; innovation; bank financing; resources; holding

IRIS IMAGES COMPARISON ALGORITHM BASED ON IRIS KEY POINTS.

• E. Pavelyeva  Faculty of Computational Mathematics and Cybernetics, M. V. LomonosovMoscow State University, paveljeva@yandex.ru
• A. Krylov   Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, kryl@cs.msu.ru

Abstract: Iris images comparison algorithm using iris key points is proposed. As the iris key points, the points with the maximal convolution values in the Hermite transform are taken. Only iris regions free of glares, eyelashes, and eyelids are analyzed to increase the effectiveness of the algorithm. The iris rotation angle estimation method is also proposed. The proposed algorithm was tested with the public iris images database CASIA-IrisV3.

Keywords:  biometrics; iris recognition; Hermite transform; iris key points

ALGORITHM OF AUTOMATIC FACE DETECTION IN THERMAL IMAGES.

• N. Basha  Institute of Applied Acoustics, IUNSM “Dubna,” natalia.basha@niipa.ru
• L. Shulga  Institute of Applied Acoustics, luda.shulga@niipa.ru

Abstract: An approach for thermal images analysis of intelligent video surveillance systems has been suggested. An automatic face detection algorithm in thermal images is proposed. In order to evaluate the performance of the algorithm, the experiments were carried out on the private database of 103 thermal images of 15 subjects obtained under different environment conditions. The experimental results reveal good performance of the proposed algorithm.

Keywords:  pattern recognition; image analysis; video surveillance; thermography; face detection

ON A REFINEMENT OF CERTAIN RESULTS FOR A BAYESIAN QUEUING MODEL.

• A.A. Kudriavtsev  Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, nubigena@hotmail.com
• S. Ya. Shorgin  IPI RAN, sshorgin@ipiran.ru

Abstract: The paper relates to an essential refinement of published earlier formula for distribution of nonloss probability in a Bayesian reliability model. The paper includes the correct formula for the probability density and improved formulae of the first two moments of mentioned distribution.

Keywords:  Bayesian approach; queueing systems; reliability; mixed distributions; modeling