ABSTRACTS OF ARTICLES OF THE JOURNAL "INFORMATION TECHNOLOGIES".
No. 2. Vol. 24. 2018

To the contents

S. P. Kovalyov, Leading Researcher, e-mail: kovalyov@nm.ru V. A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences M. Yu. Shaymardanov, Student, e-mail: mikhail.shaym@gmail.com Bauman Moscow State Technical University

Methods of Applying Manufacturing Constraints in Topology Optimization

In this paper, we formulated and solved the problem of development of topology optimization algorithm, which can be used to get optimization solutions automatically, taking into account the manufacturing constraints kind of growth directions in the cycle of engineering products design. The algorithm is obtained by the modification of the classic SIMP-method of topology optimization. We have modified the Matlab program that visualizes the optimization process in order to apply "four growth directions" constraints.
Keywords: topology optimization, SIMP-method, manufacturing constraints, growth direction method, design, engineering, algorithm, computational experiment

P. 75–80

V. D. Chertovskoy, Professor, e-mail: vdchertows@mail.ru, Admiral Makarov State University of Maritime and Inland Shipping, St. Petersburg

Methology of Mathematical Description and Modeling of Adaptive Automatized Manufacturing Control Systems

Hierarchical adaptive manufacturing control systems was researched these work in optimal regime with change goal vector composition. System analyses of mathematical description and computer realization was made. Description method and method of obtain of numerical information for model of system were proposed. Singularities were considered and program structure of computer realization, methods of numeric data agreement were chosen.
Keywords: system analyses adaptive manufacturing control system planning description modeling

P. 81–86

I. V. Lobov, Senior Researcher, e-mail: lobov@ihep.ru, V. G. Gotman, Junior Researcher, e-mail: vladislav.gotman@ihep.ru Institute for High Energy Physics, National Research Center "Kurchatov Institute", MO, Protvino

The Utilization of Ogg Multimedia Container Format for Live Streaming over HTTP Using Progressive Download Method

The general scheme of the media live streaming over HTTP has been proposed. It was shown that the general scheme has three common basic features: generalized fragments, metadata and fragment timestamp. The generalized fragment is the media container unit being transmitted over network. The fragment should contain encoded part of the media stream. The metadata stores common information (codec type, frame rate, image dimensions, bit rate, sample rate) which necessary for the client to decode and play the media stream back. The timestamp is the fragment property necessary to correct playback and synchronize for several media tracks — particularly for video and audio streams. The analysis of existing basic technologies (Apple HLS, Adobe HDS, Microsoft Smooth Streaming, MPEG DASH) of live streaming over HTTP was made. It was revealed that all these technologies meet this scheme. All of them have their own concept for the small unit of the coded media stream which matches to the proposed generalized fragment with one significant addition — the fragment has independent download possibility. The metadata and timestamps are stored into the manifest file. The technology of the media container Ogg utilization for live streaming over HTTP by progressive download method was proposed. This technology is quite different from the existing basics technologies. Yet it meets the general scheme. The comparison of the proposed technology with existing basic technologies has been made.
Keywords: live streaming, progressive download, Ogg format, Apple HLS, Adobe HDS, Microsoft Smooth Streaming, MPEG DASH

P. 87–96

L. A. Mylnikov, Associate Professor, e-mail: leonid.mylnikov@pstu.ru, A. B. Seledkova, Master Student, e-mail: aleksandraseledkova@yandex.ru, Perm National Research Polytechnic University

Selection of the Parameters Predicting Method in Production and Economic Systems Based on Risks Metrics
The article describes the choice features of methods for predicting the parameter values of production and economic systems. For this, the data features collected and used are considered, the main problem in the projects' data analysis implemented in production and economic systems will be either a small amount of statistical data or data describing a number of products, product modifications, etc. (i.e., undivided data). The main approaches used for predicting parameters are considered. These are approaches described by special curves, supervised learning, unsupervised learning and semi-supervised learning. Features of the forecasted data and also forecasts based on regression analysis, the method of support vector machine, auto regression and wavelet analysis are constructed. For the predicted values, the method of choosing the forecasting method based on the risk assessment of the introduced forecasts is given. The choice is proposed to be made on the basis of an estimate of the exact prediction for a maximum period of time. Since, starting with a certain forecast value, risk value assessment begins to increase sharply, which allow choosing the method of forecasting. Also, the application of forecasts in decision-making tasks is considered.
Keywords: Forecast, risk metrics, management parameters, regression analysis, reference vector method, wavelet analysis

C. 97–103

T. S. Osadchaya, Engineer, e-mail: taniaosadchaya6@gmail.com, A. Yu. Shcheglov, Ph. D., Professor,
ITMO University, St. Petersburg, Russia

Comprehensive Solution of Protection against Attack with Rights of Privileged Users

The work is dedicated to solvation of the problem of complex protection against attacks with privileged user rights. The following sources of this security threat are considered: legal users with privileged rights, and malicious programs running with privileged rights. The new approach of protection is described. It provides control and delineation of the actions of privileged users, including limitation of their administration capabilities, as well as excluding the possibility of the influence of malicious programs running with privileged rights on the system.
Keywords: privileged users, insiders, access control, access rights, limitation of users' administration capabilities, self-defense mechanism, malicious software

P. C. 104–109

L. A. Lyutikova, Head of the Departmen, e-mail: lylarisa@yandex.ru, E. V. Shmatova, Junior Researcher, e-mail: lenavsh@yandex.ru, Institute of Applied Mathematics and Automation of Kabardin-Balkar Scientific Centre of RAS
(IAMA KBSC RAS), Nalchik

A Logical Approach to the Correction of the Results of the Operation of SΟ-Neural Networks

The paper considers the method of constructing a logical corrector for the operation of SΟ--neural networks when solving recognition problems. A method is proposed for detecting implicit regularities, according to the structure of the SΟ-neuron, which can enhance the adaptive properties of the recognition system. It is argued that the combined approach to the organization of the recognition system increases its effectiveness and allows in cases of an incorrect response of the SΟ-neuron as a solution to indicate the objects closest to the requested attributes.
Keywords: logical analysis, data analysis, algorithm, SΟ-neuron, training sample, decision trees, corrective operations

P. 110–116

S. V. Kulikov, Researcher, e-mail: kulikov@deepmark.ru, O. S. Zakharov, Researcher, e-mail: zakharov@deepmark.ru, D. Yu. Andreev, Director, e-mail: andreev@deepmark.ru, LLC "Laboratoriya umnykh tekhnologiy", Penza, Russia

Exploring the Possibility of Using Deep Convolutional Neural Network Paired with Neural "Biometric Image to Code" Converter in Face Recognition

The article focuses on the possibility of using neural "biometric image to code" converter (NBCC) according to standards GOST R 52633.X in face-based biometric encryption. NBCC is used as last layer in pretrained deep convolutional neural network. The article describes the evaluation of indexes defined in GOST R 52633.1—2009 such as stability, uniqueness and quality of features extracted with deep convolutional neural network to analyze the possibility of training NBCC on features mentioned above. A number of NBCC configurations selected in accordance to GOST R 52633.5—2011 are evaluated and compared by ROC-curves to Euclidian distance based templates and SVMs.
Keywords: face recognition, biometric encryption, deep convolutional neural network, features, stability index, uniqueness index, quality index, support vector machine, Euclidian distance, ROC-curve

P. 116–120

S. A. Gorbatkov, D. Sc. Professor, sgorbatkov@mail.ru, Financial university under the government of the Russian Federation, Ufa (branch), Ufa 450015, Russia, D. V. Polupanov, Ph. D., Associate Professor, demetrious@mail.ru, Bashkir State University, Ufa, 450076, Russia

Optimal Selection and Aggregation of Exogenous Variables in Neural Network Models of Bankruptcies Based on Harrington Functions

The article is devoted to the issue of constructing and further improving neural network models in conditions of noise, inaccuracy, incompleteness and uncertainty of the data. A practical application is the development of an original neural network logistics method for assessing the probability of enterprise bankruptcies. In order to improve the quality, adequacy, accuracy and predictive properties of neural network models, two methods of reducing factor space are proposed: 1) an iterative method of optimal factor selection; 2) aggregation of factors using the generalized Harrington desirability function. A common original concept of the proposed methods, which differs from the known methods of exogenous variables compression, is a complex (systemic) study of the interrelation between the operations of optimal factor selection and their aggregation with the operation of regularizing the inverse problem of training neural networks in a Bayesian ensemble. The generated emergent effect is the compression of the factor space and, accordingly, facilitating the construction of a neural network model in terms of providing its required prognostic properties. The proposed theoretical ideas of the two original methods of variable compression are confirmed by the results on real data of construction industry enterprises. Quantitative estimates are presented comparing the proposed methods with the construction of neural network models on the initial set of factors, as well as with the aggregation of factors by the fuzzy matrix convolution method.
Keywords: optimal selection, aggregation of factors, neural network models of bankruptcies, system approach to modeling

P. 121–130

S. M. Avdoshin, Ph. D., Professor, Head of Software Engineering School, Faculty of Computer Science, savdoshin@hse.ru, E. Yu. Pesotskaya, Associate Professor, Faculty of Computer Science, epesotskaya@hse.ru National Research University Higher School of Economics (HSE), Moscow, 105187, Russia

Internet of Things: Transportation

The paper describes the possibility of using the Internet of Things in transport industry and processes while all the life cycle of transporting passengers and goods. Authors provide detailed research of effects and benefits by using Internet of Things while transportation, "smart cars and vehicles", customer applications that drives cost optimization and economic effect. The paper provides the analysis of potential trends in development of Internet of Things in transportation sector forced by information and digital technologies development recent years. Also the productivity and cost savings aspects are carefully examined as well as their impact to the company's transportation and logistics business processes. While providing recommendations on business processes improvement and costs and risks reducing initiatives, the global trends are considered.
Keywords: Internet of Things, mobile technologies, mobile applications, transportation, "smart cars", intellectual transport systems

P. 131–138

Le Ba Chung, Graduate Student, e-mail: chungbaumanvietnam@gmail.com, Moscow Institute of Physics and Technology (State University), Ju. A. Holopov, Leader Engineer, Lebedev Institute of Precision Mechanics and Computer Engineering

Asymmetrical Inter-Module Interface

Inter-module communications in a digital control system based on the principle of "outstretched hand" are considered in this article. A simple asymmetric inter-module interface for transferring information between peripheral devices and a computer is proposed. The interface is characterized by non-complex exchange control mechanism, simple packet structure and high data density in the packet, which allow quick "transparent" communications between remote control system devices and a central computer.
Keywords: digital control system, "outstretched hand" principle, inter-module communication, central computer, peripherals

P. 138–143

To the contents