Eng | Rus

“Systems and Means of Informatics” scientific journal

Volume 26, Issue 2, 2016

Content   Abstract and Keywords   About Authors

SYSTEMS AND MEANS OF DEEP LEARNING FOR CLASSIFICATION PROBLEMS.
  • O. Yu. Bakhteev
  • M. S. Popova
  • V. V. Strijov

Abstract: The paper provides a guidance on deep learning net construction and optimization using graphics processing unit. The paper proposes to use GPU- instances on the cloud platform Amazon Web Services. The problem of time series classification is considered. The paper proposes to use a deep learning net, i.e., a multilevel superposition of models, belonging to the following classes: restricted Boltzman machines, autoencoders, and neural nets with softmax- function in output. The proposed method was tested on a dataset containing time segments from mobile phone accelerometer. The analysis of relation between classification error, dataset size, and superposition parameter amount has been conducted.

Keywords: time series classification; deep learning; model superposition; autoencoder; restricted Boltzmann machine; cloud service

SELF-TIMING ANALYSIS OF ELECTRONIC CIRCUITS ON THE LOWER LEVEL OF HIERARCHY.
  • L. P. Plekhanov

Abstract: Self-timed circuits (independent on element's delay) have the unique properties of a lack of competitions and safe on Out-Stack-At-Fault (OSAF). They require analysis on self-timing. In the traditional approach - analyzing of elements switching, computational complexity is so great that it does not allow analyzing the most practical circuits. The functional hierarchical method, previously proposed by the author, analyzes logic equations only at the lower level, and at the upper levels, it examines only the relationships between blocks. The suggested method makes it possible to analyze circuits of any size effectively. This article describes in detail this method at the lower level of the hierarchy.

Keywords: self-timed circuits; asynchronous circuits; self-timing analysis; hierarchical analysis

LEXICAL ANALYSIS OF DYNAMICALLY GENERATED STRING EXPRESSIONS.
  • M. I. Polubelova
  • S. V. Grigorev

Abstract: There is a class of applications which utilizes the idea of string embedding of one language into another. In this approach, a host program generates string representation of clauses in some external language, which are then passed to a dedicated runtime component for analysis and execution. Despite providing better expressiveness and flexibility, this technique makes the behavior of the system less predictable, complicates maintenance, and is a source of such vulnerabilities as SQL injections and cross-site scripting. Static analysis of strings is intended to minimize the drawbacks of the approach by checking well-formedness of a set of all dynamically-generated clauses at compile-time. Lexical analysis, or tokenization, is an important step of static analysis. The paper presents an automated approach to lexical analyzers construction which simplifies implementation of static analyzers of dynamically generated code.

Keywords: string analysis; lexing; string-embedded language; lexer generator

NORMAL PUGACHEV FILTERS FOR STATE LINEAR AUTOCORRELATED STOCHASTIC SYSTEMS.
  • I. N. Sinitsyn
  • E. R. Korepanov

Abstract: The analytical synthesis theory of continuous conditionally optimal Pugachev filters for processing in linear state stochastic systems (StS) with uncorrelated and autocorrelated noises is presented. For non-Gaussian StS, first works belong to Pugachev and Sinitsyn. Basic algorithms for state linear StS with uncorrelated noises are given. Generalization of algorithms for autocorrelated state linear Sts is presented. A test example for the software tool "StS-Filter, 2016" is described in details. Some generalizations are mentioned.

Keywords: autocorreled noise; Liptser-Shiraev filter (LSF); Liptser-Shiraev conditions; normal approximation method (NAM) for a posteriori density; normal conditionally optimal Pugachev filter (NPF); stochastic systems (StS); state linear StS; statistical linearization method (SLM)

MODIFIED ELLIPSOIDAL SUBOPTIMAL FILTERS FOR NONLINEAR STOCHASTIC SYSTEMS ON MANIFOLDS.
  • I. N. Sinitsyn
  • V. I. Sinitsyn
  • E. R. Korepanov

Abstract: The analytical synthesis theory for modified ellipsoidal suboptimal filters (MESOF) for nonlinear stochastic systems on manifolds (MStS) based on the nonnormed a posteriori characteristic function is developed. Gaussian and non-Gaussian MStS are considered. Symbolic MESOF algorithms are more simple than the ESOF algorithms. The MESOF algorithms are the basis of the software tools ("StS-Filter," 2016).

Keywords: accuracy and sensitivity equations; ellipsoidal approximation method (EAM); ellipsoidal suboptimal filter (ESOF); modified ellipsoidal suboptimal filter (MESOF); nonnormed characteristic function; Poisson noise; suboptimal filter (SOF); Wiener noise

GENERATING REALISTIC STRUCTURE-FROM-MOTION DATASETS THROUGH VIRTUAL PHOTOGRAPHY.
  • O. A. Yakovlev
  • A. V. Gasilov

Abstract: The paper presents the virtual photography method of generating structure-from-motion datasets. Virtual photography is an interactive process of making shots from different viewpoints and with arbitrary camera intrinsics according to the pinhole camera model. The result of virtual photography may be considered as input data for the structure-from-motion software. The paper provides realistic datasets according to the most significant properties of real datasets. Implementation of the virtual photography method using the OpenCV library is described.

Keywords: structure from motion; synthetic datasets; virtual shooting; OpenCV

HIERARCHICAL FORMAT OF A BIOGRAPHICAL FACT.
  • I. M. Adamovich
  • O. I. Volkov

Abstract: This article focuses on the facts in the specific historical studies. The specific duality of a biographical fact, which is evidenced in its connections with the real world and with the information space, is mentioned. The hierarchical format of a biographical fact, which is used in the "T-parser" system of automatic extraction of facts from texts in natural language, is described in detail. The accordance of this format with the formal properties of a biographical fact is proven. The logic of biographical facts operators for this format is explored and described. The relation of the hierarchical format of facts with the domain ontology and the prospects of its usage in automated systems of facts processing are also demonstrated. The possibility of partial formalization at the stage of fact extraction is analyzed. The technology of biographical data processing, which automates the fact extraction stage and the stage of facts comparison with the purpose of integration and conflict resolution, including full fact formalization on the basis of an ontology, is proposed and analyzed. The problems of this approach are described and the possible ways of their solution are proposed.

Keywords: biographical fact; hierarchical format of fact; automated technology of facts processing; ontology; logic of facts

INFORMATION SYSTEM OF CELLULAR AND NEURAL NETWORK MODELING OF CHEMICAL ACCIDENT CONSEQUENCES AT DANGEROUS INDUSTRIAL FACILITIES.
  • S. P. Dudarov
  • P. L. Papaev

Abstract: An information system intended for modeling the process of dissipation of contaminant in free air as a result of emergency emission at dangerous facility is described. The information system can also solve the task of emission consequences evaluation using pollution and toxic lesion areas calculation. Cellular and neural network models are applied as mathematical means of modeling. They combine advantages of both cellular models and artificial neural networks. The architecture of the developed information system is presented by organizational and functional structures as well as by the information-logical model of its database. The stages of the software use such as data samples processing, neural network training, work with the base and generalizing models are described thoroughly.

Keywords: cellular and neural network modeling; ecological safety; chemical accident consequences; contaminant concentration; concentration area; information system modeling; software development; information system organizational structure; information system functional structure; database; information-logical model

METHODS OF CREATION OF A CIRCUITRY CAD USING THE SERVICE-ORIENTED APPROACH ON THE BASIS OF THE WEBSOCKET PROTOCOL.
  • V. I. Anisimov
  • V. N. Gridin
  • S. A. Vasilev

Abstract: The main criteria to which the web service of circuitry design shall correspond are given. The classical approach to creation of service-oriented systems is described. The article reveals the internal architecture of a web service on the basis of service-oriented architecture (SOA) and the main stages of functioning. The article describes application of SOA for automated design engineering systems, in particular circuitry CAD (computer-aided design). An example of successful implementation of the SOA architecture in automated design engineering systems is given. A number of circuitry tasks for creation of an automated design engineering system on the basis of the service-oriented approach are formulated and described. The article considers the advantages and shortcomings of use of the WebSocket full-duplex communication protocol (RFC 6455 standard) as the basis of an ESB bus of a service-oriented automated design engineering system in comparison with the SOA protocol. The advantages and shortcomings of use of the WebSocket protocol for performing external communications of a web service are given. The question of transmission of complex data types in heterogeneous software environments using the WebSocket full-duplex data protocol (RFC 6455 standard) is taken up. The advantages and shortcomings of implementation of the WebSocket communication protocol in automated design engineering systems, in particular circuitry design systems, on the basis of SOA are specified.

Keywords: SOA; CAD; web service; WebSocket; environment .NET; bus ESB; cross-platform; asynchrony; heterogeneous environment; frequency response; stationary mode; circuit engineering; difficult data types; network traffic; high-speed performance

NORMALIZATION OF ONLINE BANKING SERVICES.
  • V. D. Ilyin

Abstract: The paper presents basic problems to be solved by means of online banking services of the normalized economic mechanism. Primary online services of the central bank, banks-providers, and personal electronic banks are enumerated. More details are given for the online services of state certification of operations on bank accounts, early repayment of debts, and emission of debt amounts.

Keywords: normalized economic mechanism; central bank; bank-provider; personal electronic bank; online banking service

COGNITIVE FEATURES OF INFORMATION MODELING DURING DESIGN OF COMPLEX INFORMATION SYSTEMS.
  • G. V. Lukyanov
  • D. A. Nikishin

Abstract: The article is devoted to the problems of information modeling of complex information systems. Two of the most controversial and debatable issues of information modeling are considered: the business case for simulation and its cognitive aspects. The article shows that the "driving" force of modeling is the economic feasibility of this activity in the process of designing and creating complex and expensive systems. Cognitive aspects come to the fore during design and creation of complex information systems involving a large team of intensively interacting specialists. All studies and findings are backed up with convincing examples from the practice of design and development of complex and expensive systems, mainly, aircraft systems.

Keywords: information model; cognitive modeling; information system; design and development

INFORMATION TECHNOLOGIES AND EDUCATION IN THE EUROPEAN UNION: FORMING THE NEW PROFESSIONALS.
  • D. A. Bogdanova

Abstract: The situation in the labour market mismatch between the availability and demand for information and communications technologies specialists that exists in the European Union despite the high level of unemployment is con-sidered. The reasons for the existing imbalance that has mainly been caused by insufficient level of digital competences among the significant part of the working population are analyzed. Certain aspects of "The Riga Declaration on e-skills" designed to change the current situation are presented. A significant role in the imbalance compensation is assigned to school teachers, who either insufficiently use or avoid the ICT usage in the learning process, which has been proved by the research results. The measures taken to improve the teachers qualification level are listed. The Massive Open Online Courses focused on various aspects of the ICT use in teaching both general subjects and computer science are given as an example.

Keywords: digital competences; digital technologies; professional development; training teachers; Massive Open Online Courses; kids digital safety; selfassessment; robotics; coding





 

RUS