РУС
ENG
№ 1(36)'2025
Aims and Scope
Founder and Publisher
Editorial board
Licensing terms
Privacy Statement
Plagiarism policy
Publication ethics
Archiving Policy
Subscription
Instructions for authors
Reviewing proccess
Copyright Notice
License agreement
Article processing charges
All issues
Search
Contacts
Section
ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING
A. V. Nicenko, V. Ju. Shelepov
ABOUT THE GENITIVE AFTER A VERB IN THE CASE OF ITS HOMONYMY WITH THE ACCUSATIVE
UDC: 004.89:004.93
DOI: 10.24412/2413-7383-4-11
Language: Russian
Abstract: The article suggests a method for automatic disambiguation of genitive and accusative for a noun following the verb. The result is determined by this verb itself (the verb criterion). The method is implemented in experimental software for homonymy resolution.
Keywords: natural language processing, genitive and accusative cases after the verb, homonymy resolution.
O. I. Fedyaev, N. V. Meleshchenko
ROLE MODELS OF AGENTS OF THE SYSTEM FOR MODELING THE PROCESS OF UPDATING ACADEMIC DISCIPLINES IN ACCORDANCE WITH THE REQUIREMENTS OF ENTERPRISES
UDC: 004.853
DOI: 10.24412/2413-7383-12-25
Language: Russian
Abstract: The process of extracting new competencies from the texts of recommendations of enterprises to university graduates has been formalized. This will make it possible to update the curricula of the disciplines of the graduating department of the university in a timely manner, taking into account the requirements of the labor market. The task was solved by computer processing o f the texts of recommendations in natural language using machine learning methods. The algorithm for its solution implements a special software agent with a BDI architecture in interaction with other agents that simulate the roles of enterprises and teache rs based on the principle of limited rationality. Experimental studies of the developed algorithms and programs have been carried out.
Keywords: university department, requirements of enterprises, academic programs of disciplines, knowledge extraction from text, machine learning, software agents.
Section
SYSTEM ANALYSIS, CONTROL AND INFORMATION PROCESSING, STATISTICS
Голобоков А.М., Коханов А.А., Кац И.Д., Мутина Е.И., Адамова Л.Е
КОМПЛЕКСНАЯ СИСТЕМА ИИ НА ОСНОВЕ МИВАРНЫХ И НЕЙРОННЫХ СЕТЕЙ ДЛЯ ОБНАРУЖЕНИЯ ДЕФЕКТОВ ПРОДУКЦИИ НА ПРОИЗВОДСТВЕ ХЛЕБОБУЛОЧНЫХ ИЗДЕЛИЙ
UDC: 004.891+ 007.52 + 004.896 + 681.518 + 65.011.56
DOI: 10.24412/2413-7383-27-42
Language: Russian
Abstract: Golobokov A.M. 1, Kokhanov A.A.1, Katz I.D.1, Mutina E.I. 2, Adamova L.E.3 1Institute of Artificial Intelligence of the Russian Technological University MIREA 119454, Vernadsky ave., bld. 78, Moscow, Russia 2 JSC M. A. Kartsev Research Institute of Computing Systems, 117437, Profsoyuznaya st., bld. 108, Moscow, Russia 3Russian New University (RosNOU), 105005, st. Radio, 22, Moscow, Russia COMPLEX AI SYSTEM BASED ON MIVAR AND NEURAL NETWORKS FOR DETECTION OF PRODUCT DEFECTS IN BAKERY PRODUCTS PRODUCTION The production of bakery products includes many stages, from the preparation of raw materials to the packaging of finished products. One of the key tasks in this area is to ensure product quality control. Possible violations in the production environment of bakery products are analyzed, a system of factors influencing them is established, and a corresponding forecasting model is built. It is substantiated that in the field of smart production systems for the production of bakery products, it is possible and advisable to create a mivar expert system to increase the intellectualization of decision- making and information processing. A mivar knowl edge base has been developed, including 54 mivar "If, Then" rules, to detect product defects in the production of bakery products. The potential of new opportunities for further implementation of the mivar approach in combination with machine learning algorithms and neural networks at various stages of production is substantiated. This will improve the quality of manufactured products and move to a new level of creating automated control systems for production systems in the field of bakery production.
Keywords: mivar approach, artificial intelligence, defect detection, Wi!Mi RAZUMATOR, mivar technologies, mivar expert systems, machine learning, bakery products, machine vision.
3JSC M. A. Kartsev Research Institute of Computing Systems
MIVAR EXPERT SYSTEM FOR INTELLECTUALIZATION OF AUTOMOBILE WHEEL DISKS PRODUCTION
UDC: 004.891+ 007.52 + 004.896 + 681.518 + 65.011.56
DOI: 10.24412/2413-7383-43-57
Language: Russian
Abstract: решений, производство автомобильных колесных дисков. It is substantiated that in the field of smart production systems for the production of wheel rims, it is possible and advisable to create a mivar expert system (MES). A system analysis of the subject area was performed, a formalized description of the subject area was developed, a mivar knowledge base was created in the form of a tabular representation of 66 rules for the mivar network for describing decision- making processes and information processing in the production of wheel rims. The mivar expert system has been implemented in the special mathematical software Wi!Mi Razumator version 2.1. Successful testing of mivar models of the subject area of smart production systems for the production of wheel rims has been carried out.
Keywords: mivar, artificial intelligence, smart production systems, mivar expert systems, Wi!Mi Razumator, mivar technologies, system analysis, information processing, management, decision making, production of automobile wheel disks.
K. S. Ivashko, S. A. Izosimova, V. N. Piguz, К. V. Kovaleva
THEORETICAL ASPECTS OF DIAGNOSTICS AND SELF-REGULATION OF PSYCHOEMOTIONAL STATES OF PERSONALITY
UDC: 004:159. 923.5
DOI: 10.24412/2413-7383-58-67
Language: Russian
Abstract: The article analyzes the existing methods in the diagnosis of psychoemotional states, considers various drug -free app roaches to improve self -regulation, such as psyc hotherapy, meditation and relaxation techniques, as well as the theoretical basis, features and possib ilities of application of these methods in everyday life. The factors influencing the effectiveness of the se methods, including individual personality cha racteristics, the influence of stress and social environment, are determined. Theoretical analysis of scientific literature is carried out, empirical data are systematized and generalized.
Keywords: self-regulation of psychoemotional states of personality, research of theoretical aspects of diagnostics, cognitive reappraisal and behavioral activation
R. V. Kovalchik
DEVELOPMENT OF MACHINE LEARNING MODEL OF PREDICTION HIGH-TEMPERATURE METALLURGICAL PROPERTIES OF COKE
UDC: 004.932.2:004.622
DOI: 10.24412/2413-7383-68-78
Language: Russian
Abstract: The results of coke high -temperature metallurgical properties prediction models developed by the author are presented. The optimal model of machine learning was selected, which allowed to achieve the highest accuracy of prediction of the target index, as which the coke reactivity index CRI is used. The peculiarity of the models is the use of coal charge grade composition as input values. The proposed approach does not require additional laboratory studies of physicochemical properties of coal charge for coke production.
Keywords: machine learning, neural networks, optimizers, metallurgical coke, forecasting.
M. V. Minkovskaya
DEVELOPMENT OF THE DIGITAL POLICY OF THE EAEU COUNTRIES AND RUSSIA IN MODERN CONDITIONS
UDC: 334.021
DOI: 10.24412/2413-7383-79-92
Language: Russian
Abstract: An analysis of digitalization processes in the EAEU was carried out, an emphasis was placed on a cycle transformation of business development and the transformation paths, taking into account the national interests of the countries of the Union. The main elements of the digital policy of the Union have been allocated, the factors affecting the pace of civilization of countries are indicated. Assessment of the risk group of digital transformation of the EAEU and the evaluation and forecasting model that allow you to effectively implement the digitalization process are indicated.
Keywords: digital policy, transformation, digitalization models, digitalization level, economic effect, regional e-commerce, digital agenda.
Section
МЕТОДЫ И СИСТЕМЫ ЗАЩИТЫ ИНФОРМАЦИИ, ИНФОРМАЦИОННАЯ БЕЗОПАСНОСТЬ
М. В. Бабичева, И. А. Третьяков
валась генерати вная нейронная сеть (GAN), а для распознавания – сверточная (CNN)
UDC: 004.932.2
DOI: 10.24412/2413-7383-94-105
Language: Russian
Abstract: сеть, Error Level Analysis Deepfakes are often used as tools for committing crimes against individuals and states, necessitating technical means to determine the artificial origin of images. This study explores various approaches to detecting fake images, including texture analysis, neural networks, and anomaly detection algorithms. The goal of the research is to investigate methods of creating fake images, identify their distinguishing features, and develop a d etection methodology using deep learning and neural networks. For generating deepfakes, a generative adversarial network (GAN) was employed, while a convolutional neural network (CNN) was used for recognition. The proposed model achieves 89% accuracy in detecting fake images, performing on par with most foreign counterparts.
Keywords: deepfakes, generative adversarial network (GAN), convolutional neural network (CNN), Error Level Analysis
Y. Imamverdiyev, E. Baghirov, I.J. Chukwu
IMAGE-BASED DEEP LEARNING METHOD FOR EFFECTIVE MALWARE DETECTION
UDC: 519.71
DOI: 10.24412/2413-7383- 106-122
Language: Russian
Abstract: В статье рассматривается метод обнаружения вредоносного программного обеспечения на основе анализа изображений, представленных в оттенках серого. Для анализа используются 13 современных сверточных нейронных сетей, включая DenseNet201, MobileNet, и другие, на основе набора данных Malimg. Проведены эксперименты, включающие обучение и настройку гиперпараметров для оптимизации производительности моделей. Показано, что модели, такие как DenseNet201 и MobileNet, достигают высокой точности, полноты, точности и F1-метрики. Данный подход позволяет улучшить процесс обнаружения вредоносного ПО, обеспечивая высокую эффективность и устойчивость к традиционным методам обхода систем защиты. Область применения работы — современные системы кибербезопасности, включая разработку новых методов анализа вредоносного ПО и защиту от кибератак.
Keywords: Malware detection, malware analysis, transfer learning, image-based detection, cybersecurity
V.R. Romaniuk
PRINCIPLES OF DEVELOPING A SYSTEM FOR DETERMINING EYE MOVEMENT ACTIVITY BASED ON DATA FROM A MOBILE PORTABLE ELECTROENCEPHALOGRAPH
UDC: 004.8
DOI: 10.24412/2413-7383-123-133
Language: Russian
Abstract: Eye movements play an important role in human cognitive processes, making them a subject of interest across a wide range of scientific and applied fields. Traditional eye-tracking methods offer high accuracy but have limitations related to the use of cameras or their equivalents. This paper proposes principles for developing a system to detect eye movement activity based on data from mobile, portable electro - encephalograph (EEG), which allows overcome these limitations. The system design is based on machine learning methods such as Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNN). Key aspects of development include signal preprocessing, filtering, data normalization, and feature extraction. The proposed principles lay the foundation for developing eye movement detection systems applicable in natural environments.
Keywords: EEG, eye movement activity, machine learning
Section
ROBOTS, MECHATRONICS AND ROBOTIC SYSTEMS
V. M. Zuev
ORTHOGONALIZATION NEURAL NETWORK PONDUS COEFFICIENTIUM
UDC: 004.93
DOI: 10.24412/2413-7383-135-140
Language: Russian
Abstract: The article considers the regularization of a convolutional neural network using the orthogonalization method. The proposed method can be used to reduce the number of optimized weight coefficients. A preliminary filter adjustment stage is introduced, which allows choosing the optimal number of convolution layers and speeding up training due to the optimal filter adjustment at the beginning of training. The analysis shows that the method we propose is quite efficient.
Keywords: neural network, regularization, orthogonalization. Articulus disserit regu larizationem retis neuralis convolutionis utendi methodo orthogonalizationis. Proposita methodus adhiberi potest ad reducere numerum coefficientium optimized ponderis. Filtrum
A. L. Okhotnikov1, A.V. Zazhigalkin2
OVERVIEW OF KEY TECHNOLOGIES OF ROBOTECHNICS
UDC: 001.895; 621.865.8, 629.066
DOI: 10.24412/2413-7383-141-155
Language: Russian
Abstract: The article describes the applied technologies for the development of robots and robotics, including artificial intelligence. The current state of domestic robotics is assessed. Promising areas of work on robotisation of production processes in JSC ‘Russia n Railways’ are listed. Modern algorithms and models of sensor data processing and requirements to convolutional neural networks (CNN) for vision systems are analysed. The directions of prospective research in the field of development of robotic systems and complexes in the railway sector are proposed.
Keywords: robotic complex, biomorphic robot, artificial intelligence, vision system, predictive analytics, convolutional neural networks.
