For the generalized elastic-plastic material of Murnaghan the software implementation procedures were obtained and a set of programs was developed to calculate the criterion deviator and the external normal vector to the surface of the deviator section of the yield surface. The criterion deviator is decomposed into the sum of three deviators as isotropic, anisotropy and the vortex. Anisotropy deviator is formed from anisotropic structures of the second and third degrees of the specific potential energy of elastic deformation. The transition from material indifferent quantities into the expansion to invariant quantities is made by orthogonal transformation with the orthogonal elastic rotation tensor itself. The necessary analytical relations were obtained for calculating the transformed criteria deviator as a symmetric operator in the five-dimensional vector space of stress deviators. The vectors of external normal for the singular point of the yield surface in the space of stress deviators are determined by two eigenvectors of the operator, and for a regular point by one of them. The calculation of quantities sought is carried out for the elastic type of Murnaghan material of general form – triclinic.

In radar systems angle information about an object is obtained by discriminator, the basic properties of which are determined by the discriminator curve. Its main parameters are the conversion factor and the width. Currently, these parameters remain unguided during the tracking process. At the same time, when changing modes or conditions of operation, it is desirable to control these parameters, which is possible in multipath radars. In multipath radar the discriminator curve is formed by simultaneous weight summing of the signals at the receiving channels outputs. The objective of the study was to develop a method for finding the values of optimal weight coefficients for obtaining the desired parameters of the discriminator curve with an instantaneous amplitude comparison of signals. As an optimality criterion, we used the condition of minimizing of noise variance at the discriminator output, taking into account linear constraints on its characteristics parameters. The solution is carried out by method of Lagrange multipliers. Two options of the discriminator curve synthesis are considered: for additive and multiplicative discriminators. Expressions for optimal weight coefficients are obtained in general term. The proposed method allows to find the desired parameters of discriminator curves, in particular, it is possible to form its zero values in angular directions to suppress the influence of external noise.

Unreliable queuing systems are of considerable interest both in mathematical terms and for applications. Systems with stationary Poisson flows of customers and breakdowns and exponentially distributed service and repair times are mainly considered. This circumstance greatly simplifies the mathematical analysis of the corresponding models but rarely occurs in real systems, especially in telecommunications networks. The purpose of this study is to analyze the stationary behavior of a multi-server unreliable queueing system with a batch Markovian arrival process, which takes into account the correlation and bursty nature of real traffic. The service and repair processes are described by phase type distributions which makes it possible to take into account not only the average service and repair times but also the variance of these times. As a result of the research, the operation of the system is described by a multi-dimensional Markov chain. The condition of ergodicity of this chain is presented in a simple algorithmic form. An algorithm for calculating the stationary distribution is proposed. Formulas for the key performance characteristics of the system are obtained in terms of the stationary distribution of the Markov chain describing the system dynamics. The results can be used to make expert decisions in analyzing the performance and design of various telecommunication networks.

A mathematical model is proposed for the processes of websites scanning and addressing (selective) distribution of the text information found after the request of users in the form of their information profiles, i. e. accumulated archives of relevant Internet publications. The functional components of such an information system are three subsystems: the subsystem of texts indexing, user archives and short messages; web page scanning subsystem; the subsystem of address distribution of texts and short messages to users. Indexing the texts, archives of users and short messages is reduced to the construction of their verbal-associative networks. The indexing subsystem includes a set of linguistic dictionaries for calculating the informational content of words and verbal-associative links between them. Dictionaries are formed based on the use of publications from user’s archives. Scanning the web pages is carried out on the basis of software solutions in the form of specialized agents, whose main task is to obtain systematically and accumulate new data from updated pages. Scanning is implemented as the sequence determined by a special ordering relationship, which is fixed on the set of web pages of each web site scanned. Distribution of found publications occurs by comparing the verbal-associative networks of these publications and users’ information profiles in the form of search images.

The problems of designing modern VLSI and SoC are analyzed. The most difficult problems of design are problems of verification of projects at different stages of design. Along with the problems that arise and are solved in the mode of favorable design, in the last decade there was a problem of protection and additional control of projects in order to detect unauthorized third-party interference in the project with different fundamental goals. We consider the formation of a common approach to solving problems of control and verification in the design of modern integrated circuits based on the analysis of fault models of structural realizations of digital devices, errors arising in the design process, as well as deliberate distortions during the design and manufacturing stages, i. e. creation and development of taxonomy of possible deviations in the project. The algorithm of logical obfuscation and coding of digital structures based on the use of methods and means of test diagnostics is proposed.

The research results of the methods for formation of separable sum codes with the minimum number of undetectable errors in data vectors are presented. A formula for counting the number of undetectable errors in data vectors and codes family properties are given. A universal method for formation of such codes is shown, which makes it possible for each value of the data vector length to obtain a whole family of codes that also have different distributions of undetectable errors by type and multiplicity. An example of codes formation, methods for analyzing characteristics, code comparison are presented. A method for synthesizing coders of developed sum codes is suggested. 

The method of classification of erythrocytes (red blood cells) based on spectral features of the cell surface images (of physical-mechanical properties maps) obtained with an atomic-force microscope (AFM) is proposed. Each scan line of the original AFM image is considered as a random sequence realization and the discrete Fourier transform is applied to compute its spectral features. The spectral estimates are smoothed on the map and the informative characteristics are computed as the medians of the spectrogram values for each frequency. The classification or two classes of erythrocytes (spherocytes and discocytes) taken from patients with hereditary spherocytosis was carried out by the obtained informative characteristics using the decision trees and boosted decision trees methods. The frequency interval was found with the best classification accuracy – over 82 % for the boosted decision trees method.

This paper addresses the problem of dependence of the success rate of adversarial attacks to the deep neural networks on the biomedical image type and control parameters of generation of adversarial examples. With this work we are going to contribute towards accumulation of experimental results on adversarial attacks for the community dealing with biomedical images. The white-box Projected Gradient Descent attacks were examined based on 8 classification tasks and 13 image datasets containing more than 900 000 chest X-ray and histology images of malignant tumors. An increase of the amplitude and the number of iterations of adversarial perturbations in generating malicious adversarial images leads to a growth of the fraction of successful attacks for the majority of image types examined in this study. Histology images tend to be less sensitive to the growth of amplitude of adversarial perturbations. It was found that the success of attacks was dropping dramatically when the original confidence of predicting image class exceeded 0,95.

The problem of finding local extrema on halftone images is considered. Well-known block-search algorithms provide high speed, but they extract only strict (single-pixel) extremes, skipping extreme areas formed by non-strict extremes. Morphological search algorithms provide the selection of non-strict extremes, but have a high computational complexity. A mathematical model and an algorithm based on the brightness analysis of adjacent homogeneous regions are proposed to isolate strict and non-strict local extremes of images with low computational complexity. Their differences from well-known models are: consideration of homogeneous areas, which are formed by non-strict extremes and are local maxima or minima in relation to adjacent areas; elimination of iterative processing of non-extreme pixels; assigning the numbers to local extremes during their search. These differences allowed to increase the accuracy of local extremum extraction in comparison with block search and to reduce the computational complexity in comparison with morphological search. 

The description of the technology of modern natural language interfaces for intelligent systems and language interfaces for question-answering intelligent systems is presented, as well as methods and principles for their design. Analysis of the intelligent systems with natural language interfaces used in different areas are given. These areas are medicine, smart home technology, education, industry, fast adaptation to new technologies. The list of the most popular services with the natural language interfaces is presented. Each service can be used as a ready-to-use personal assistant or as a core for the development of a new customized natural language interface. The research of the natural language interfaces was conducted from the point of view of the natural language usage for the interaction between a user and the machine. The main problems here are the bias in natural language and the difficulties in the design of natural language interfaces that meet user expectations. The main principles of modeling of natural language interfaces are considered. As an intelligent system the interface consists of the database, knowledge machine and user interface. Speech recognition and speech synthesis components make natural language interfaces more convenient from the point of view of usability.