ISSN 0236-235X (P)
ISSN 2311-2735 (E)

Bookmark

Next issue

3
Publication date:
16 September 2019
-->

Journal articles №2 2019

11. An improved neural network training algorithm [№2 за 2019 год]
Authors: V.N. Zuev (zvn_tver@mail.ru) - R&D Institute Centerprogramsystem, Tver State Technical University, chair of Information’s Systems (Head of Laboratory), ; V.K. Kemaykin (vk-kem@mail.ru) - Tver State Technical University, Information Systems Department (Associate Professor), Ph.D;
Abstract: The paper describes heuristic modification of backpropagation algorithm using for remote batching. The backpropagation algorithm is a common algorithm for neural network training. It causes some difficulties. The main problem is enabling the generalizing of a neural network. The ability of general-izing is a most important characteristics of a neural network. It assumes that a neural produces antici-pated values on data that is not a part of a training process. However, using of noisy data causes re-training and decreasing of a generalizing ability of a neural network. The problems considered in the paper are an important part of a neural network training process. The paper describes a method to use objective functions effectively. The proposed method allows more effective calculating of the values of a goal function that is a base of the backpropagation algorithm. It also ignores failure values in training data and excludes them at earlier stages. In addition, the method allows using heterogeneous data samples for training neural networks, as well as taking into account prior information on the significance of some examples when training. The paper describes the algorithm of the proposed method. The method will improve the accuracy of a neural network for classification and regression tasks.
Keywords: neural network, backpropagatio, objective function, error function, learning algorithm
Visitors: 110

12. Comparison of neural network models for classifying text fragments containing biographical information [№2 за 2019 год]
Authors: Glazkova A.V. (a.v.glazkova@utmn.ru) - Tyumen State University, Ph.D;
Abstract: The paper presents the results of comparison of neural network architectures to solve the automatic text classification problem. The paper focuses on the topic classification of sentences containing bio-graphical facts. The data set was collected on the basis of Wikipedia and represented as a corpus of bi-ographical texts with meta tagging. Each sentence in the corpus is marked according to whether it con-tains any biographical information and what type of information it contains. The paper describes the results of topic classification of the corpus sentences for five neural net-work architectures. They are: feedforward network (FNN), recurrent neural network (RNN), gated re-current unit RNN (GRU), long short-term memory RNN (LSTM) and bidirectional LSTM (BLSTM). These models were trained on the word embedding features obtained for the corpus sentences words. We have tried two types of classification: binary (depending on whether a sentence contain biograph-ical information or not) and multi-class classification (definition of the type of biographical fact). LSTM and BLSTM models have shown the best results. The obtained results will be used to implement biographical information retrieval methods for natural language texts.
Keywords: text categorization, neural network, natural language processing, recurrent neural network, biographical fact
Visitors: 170

13. Neural network user behavior analysis for detecting host-level intrusion [№2 за 2019 год]
Authors: V.N. Zuev (zvn_tver@mail.ru) - R&D Institute Centerprogramsystem, Tver State Technical University, chair of Information’s Systems (Head of Laboratory), ; Efimov A.Yu. (efimovay@cps.tver.ru) - R&D Institute Centerprogramsystem (Head of Department);
Abstract: The paper focuses on applying machine learning for detecting anomalies in user behavior. The number of known attacks is rapidly increasing every year. In order to resist that treatment, there is a need for effective security systems, such as Intrusion Detection Systems (IDS). This type of sys-tems usually uses signature analysis and requires signature updates. Such systems are not capable of detecting unknown attacks. The methods based on anomaly detection are more attractive as they can identify previously un-known attacks without preliminary creating of intrusion signatures for every possible attack. One of the most popular directions for host-based IDS in anomaly detection is user behavior analysis. The paper describes a method of detecting user behavior analysis anomalies based on artificial neu-ral networks. For detecting user behavior anomalies, the method uses the information on user’s com-mands extracted from system log files and software. This information is converted into time series that is used to forecast next user’s commands. The number of forecasting errors determine the presence of an anomaly in user behavior. The experimental results demonstrate that the proposed method is good for detecting anomalies in user behavior, and has low probability of false positive.
Keywords: intrusion detection, computer attack, event analysis, user behavio, anomaly detection, neural network, machine learning, pc "rebus-sov"
Visitors: 185

14. Efficiency evaluation of methodical and algorithmic support of a special purpose decision support system [№2 за 2019 год]
Authors: Dopira R.V. (rvdopira@yandex.ru) - NPO RusBITex (Professor, Head of Department), Ph.D; A.V. Getmanchuk (getmanshuk@mail.ru) - Taganrog Research Institute of Communication (Deputy Head of Laboratory), Ph.D; Potapov A.N. (potapov_il@mail.ru) - Military scholastic-scientific centre of the Air forces "Air forces academy named by prof. N.E. Zhukovsky and Yu.A. Gagarin", Ph.D; M.V. Semin (potapov_il@mail.ru) - Military Scholastic-Scientific Centre of the Air Forces "Zhukovsky and Gagarin Air Forces Academy" (Head of Laboratory); D.Yu. Brezhnev (dimanbreg@mail.ru) - Military Academy of the Aerospace Defence (Doctoral Student), Ph.D;
Abstract: To eliminate information overload when assessing the state of air and space radio control objects for decision-making, we use methodological support and classification algorithms of decision support system (DSS) objects of an automated control system (ACS) for special purposes. The paper presents a method for efficiency evaluation of algorithmic support of a special purpose ACS DSS. The method is based on artificial generation of input data, as well as data representing a set of all possible values based on an exhaustive enumeration in the form of generating programs of ran-dom objects and random areas. This approach uses the analysis of the relative location of regions in a parametric space, which is a catalog of reference values. In this approach, there is no need to compare object parameters with the parameters of all catalog classes. After finding the first input into an area, we consider the possibility of getting the analyzed object into those areas that have intersections with the area in which the object has already got. The obtained experimental results of evaluation of a general computing labor intensity of the spe-cial-purpose ACS DSS when classifying the objects using a combined method is, on average, two times lower than the basic classification method. The obtained result is particularly relevant in the classifi-cation of air and space radio engineering control objects under conditions of intensive real-time data entry.
Keyword:
Visitors: 89

15. An information technology for verifying special software of military automated systems [№2 за 2019 год]
Authors: R.E. Zhidkov (r-zhidkov@rambler.ru) - Military Academy of the Aerospace Defence (Postgraduate Student); D.S. Viktorov (victorov.dmitry@yandex.ru) - Military Academy of the Aerospace Defence (Associate Professor), Ph.D; E.N. Zhidkov (eng1964@mail.ru) - Military Academy of the Aerospace Defence, Ph.D;
Abstract: There is a need to create an information technology of military automated system special software verification that allows effective checking of functionality requirements. This is due to the need to create high-quality software within time and financial constraints of the development process. The paper proposes an information technology that is a combination of methods and software utilities for pro-cessing a source code of the investigated program and technical documentation in order to obtain information of functional defects. The described information technology might be used both at the coding stage and at different testing stages in addition to the existing solutions for searching functionality defects. The verification method has a structure determined by the requirements of regulatory documents in the soft-ware development. This method includes a static analysis technique and a technique for assessing analysis com-pleteness based on source code characteristics. This verification method is designed to check the correctness of the implementation of information-calculated tasks in automated systems for managing troops and military equipment based on the principle of dimensional homogeneity of physical equations. The verification information technology is described as an IDEF0 functional model detailed to the second nesting level with a description of inputs and outputs of all works. In order to support the verification process, the authors propose software utilities for automating routine operations: generating unique identifiers of software ob-jects, searching and intentional defects introducing. These software tools are based on the infrastructure of LLVM compiler and Clang frontend that allow checking the program codes in C, C++, Objective-C.
Keywords: information technology, verification, software, idef0, functional defects, static analysis
Visitors: 187

16. A complex mathematical model of the air anti-stealth airship radar system [№2 за 2019 год]
Author: S.V. Susha (sushasv2008@yandex.ru) - Department of Ministry of Defense of Russia (Head of Department);
Abstract: One of the stages of creating an anti-stealth airship radar system and evaluating its effectiveness is the development of an integrated mathematical model. This paper presents the structure of such model. The model is developed to substantiate the technical appearance of the system and its application features, as well as to evaluate operating effect and combat (information) capabilities. A mathematical model of the system is based on the basic construction principles and a complex structure. The system itself is an information tool representing a complex radio system. The hardware and software of the system should be located both on the ground and in the air. The structure of the system mathematical model includes several functionally complete blocks and models, in particular, models of a target environment, the Earth, on-board systems of the airship radar system, etc. The paper describes particular mathematical models that are a part of an integrated one, their mutu-al interaction. The radar model of the dirigible radar system is based on calculating detection parameters (signal-noise ratio) using the radar equation. It is an analytical model of the receiving-transmitting path and a signal propagation process. Successive changing of its state at certain time intervals simulates the func-tioning dynamics. The model provides a dynamic playout of the input information to the similar output information of a simulated tool. In addition, it provides the ability to simulate an exchange with a con-trol system, has parameters that allow simulating the control of simulated device operation modes. The paper considers the possibilities of a complex mathematical model of an anti-stealth airship ra-dar system.
Keywords: airship radar system, radar station, stealth air target, complex mathematical model
Visitors: 146

17. Problems and methods of their decision in the net automatical systems of the Navy's technical support control [№2 за 2019 год]
Authors: E.I. Mukhitov (mukh70@yandex.ru) - Management of the Baltic fleet (Сhief of the Bureau of invention and rationalization); A.A. Bavula (bavel_@mail.ru) - “Soviet Union Fleet Admiral N.G. Kuznetsov Naval Academy” of the Navy Military Educational and Scientific Centre (Kaliningrad branch) (captain 2st rank, adjunct);
Abstract: The object of investigation is a net structure of the Navy maintenance support. The development of lo-gistical intelligent systems, virtual agencies of military control, net strategy, and other advanced tech-nologies will be based on experience, knowledge, intuition of officer managers and will be integrated into artificial intellect systems for military purposes. The net architecture will contribute to creating of intellectual environment in the socio-, techno- and cyber spheres, all types of human activities, includ-ing the further work of net operators, who are able to operate information, integrate into knowledge (scenarios) and provide ready-made solution options for the officials of the maintenance support con-trol agencies using artificial intelligence. One of ways of creating a highly effective net architecture is algorithmisation and growth of net computing resource accessibility. Modern and practical accents are moving from the “computer” con-cept to the “net” concept, and from intelligent activities of officer operators of military control agen-cies to introduction of virtual net operators, who are able to process badly formalised information (to aggregate, separate, refine, integrate, analyze, etc.). Therefore, automated control systems are able to provide mathematically adjusted ready-made scenarios of decisions for decision makers using software and specializing artificial intelligence systems. The novelty of the proposed method is that during decision-making, as far as the hierarchy of the maintenance support control increases, the comparative analysis and the criterial estimation in a quali-ty spiral occur simultaneously. The integration of these processes is an additional important factor for modernization of advanced automated control systems. Modern wars and their transition into the cy-berspace do not leave the time for the important decision-making stages for the officials of the control agencies. Therefore, the proposed scenario-logistical approach and the algorithmisation of mainte-nance support processes will give the opportunity to reduce the time and provide a sufficiently high quality level of administrative decisions.
Keywords: artificial intelligence, net operator, cyberspace, technical support, control agency, integration, decision, navy
Visitors: 153

18. A set-theoretic model of a functional approach to intellectualization of building and construction management processes [№2 за 2019 год]
Author: R.V. Dushkin (roman.dushkin@gmail.com) - VoiceLink Ltd (Deputy Director General on Intelligent Transportation Systems and a hardware-software package Smart City);
Abstract: The paper describes a functional approach to managing the internal environment of an intelligent building. Such approach implies using and transferring unchangeable states between the peripheral equipment and central engineering systems, the absence of side effects during control actions, as well as the construction of a decentralized network of computations and decision making on an end-user device side based on the Internet of things. This allows simultaneous obtaining of all benefits of differ-ent consideration paradigms of management processes and intellectualization. Additionally, there is an possibility of emergent showing new features of the general approach to increase the degree of control-lability and the operation efficiency of controlled objects. The paper considers mathematical foundations of the organization of a distributed computing envi-ronment for implementation of a functional approach. It also briefly describes possible control scenar-ios in various operation modes of engineering systems of intelligent buildings. The authors describe a set-theoretic model of an intellectualized building management system. This model is a theoretical ba-sis for recommendations on changing the paradigm of constructing a general automated control system for an intelligent building from the automation of engineering processes to the management intellectu-alization and the creation of an artificial intelligence system that implements a complete and autono-mous cycle of building management. The functional approach complete with new technologies (artificial intelligence, decentralized da-tabases, the Internet of things) for managing such objects as intelligent buildings, allows transferring the operation of such buildings to a higher level of service availability, sustainability, environmental friendliness and comprehensive development of not only a controlled object, but also the hierarchy of its supersystems (municipality, region, state).
Keywords: functional approach, intelligent building, Internet of things, decentralization, artificial intelligence, intellectualization
Visitors: 148

19. Programs for simulating temperature fields in cylindrical products [№2 за 2019 год]
Authors: Margolis B.I. (borismargolis@yandex.ru) - Tver State Technical University, Ph.D; Mansoor Gubran Ali (gubran_ali@mail.ru) - Tver State Technical University, Department "Automation of technological processes";
Abstract: The paper describes a temperature field model in a cylindrical product during an asymmetric convective-radiative heat transfer of surfaces with the environment and protecting surfaces (heating elements) of production equip-ment. The equations for calculating temperature distribution in the product using numerical finite-difference methods. There is a calculation example of convective-radiative cooling of a cylindrical body. It is stated that the described problem is possible to solve in Matlab. Based on the standard MatLab function pdepe, there is a developed program that allows simulating a product temperature field according to specified on thermo-physical characteristics of the material (thermal conductivity, temperature conductivity) and parameters of the convective-radiative heat transfer (a convective heat transfer co-efficient and given degrees of blackness). The paper describes the program development features related to assign-ing functions of differential equations, initial and boundary conditions. There are the main functions of the devel-oped programs, as well as the temperature distribution calculation results. The authors present a comparative analysis of the problem solution in Matlab with the results obtained using finite-difference relations. They also demonstrate their good agreement. A two-dimensional heat conduction prob-lem in a cylindrical body is solved using a parabolic function, which is one of Matlab tools to solve parabolic type problems. The cylindrical domain has been triangulated using the initmesh function. There are some fragments of the program codes of boundary conditions functions and a visual image of a two-dimensional temperature distri-bution in a hollow cylinder. When solving a thermal two-dimensional cylindrical problem, the temperature gradi-ents in height are almost absent. Therefore, there is an excellent coincidence of the temperature field with the re-sults obtained by finite-difference methods and using the function pdepe. The developed methods in the long term allow solving problems of modeling temperature field in high-quality cylindrical glass products (a glass, a bottle, a flask) that have a side face and a back surface.
Keywords: two-dimensional heat conduction problem, cylindrical product, convective-radiant heat transfer, finite difference methods, modeling, initial and boundary conditions, differential equation, finite element method, triangulation, a tempering
Visitors: 192

20. Special aspects of DWDM technology for fiber optic multiplexing [№2 за 2019 год]
Authors: Yu.M. Lisetskiy (Iurii.Lisetskyi@snt.ua) - S&T Ukraine Company (Director General), Ph.D; Yu.S. Perekopayko (Yuri.Perekopayiko@snt.ua) - S&T Ukraine (Senior Consultant);
Abstract: The paper considers the importance of DWDM technology for wavelength division multiplexing of fi-ber optic channels that enables multiplication of bandwidth of a fiber optic cable infrastructure. It is achieved by transmitting of up to 160 independent information channels through one optic fiber. Dif-ferent wavelengths are used to carry traffic with different protocols and speeds. This is an important problem for telecommunications service providers owning numerous fiber optic backbone networks and facing the need to dense the traffic in view of dynamic growth of the amount of subscribers, aver-age duration of communication sessions and introduction of new non-voice services. Conventional fiber optic technologies use one pair of fibers to transmit one channel data. On the contrary, the basic feature of DWDM is the ability to transmit N channels through one pair of fibers. It is achieved through transmitting each data stream on adjoining carrying frequencies. Their operating ranges in DWDM technology are usually called wavelengths. The paper analyzes optic multiplexers combining wavelengths into one composite signal before en-tering the fiber optic line. The receiving side then decomposes it into different channels. Their essential feature is the ability to decompose and compose composite signal to single out a wavelength or set of wavelengths and transmit the rest further down the line unchanged. There are demonstrated features of optic multiplexers as well as the evolution of their development. The paper describes the conceptions of Colorless & Omnidirectional ROADM which eliminates the need to reserve a wavelength on a permanent basis or automatically reroute services through vacant wavelengths on another direction in case of line failure. It is noted that the concepts of omnidirectional and colorful DWDM together with Flex Spectrum technology and dynamic plane of WSON manage-ment are key elements of innovative solution for Cisco nLight fiber optic transport networks. Typical-ly, the DWDM technology utilizing ROADM nodes is used to modernize or expand existing fiber optic networks to increase their bandwidth and availability.
Keywords: nodes, wavelengths, multiplexing, multiplexers, transport networks, fiber optic channels, dwdm technology
Visitors: 119

← Preview | 1 | 2 | 3 | Next →