For authors
Search
Search Results
##search.searchResults.foundPlural##
-
MODERN METHODS OF HYPERSPECTRAL IMAGE PROCESSING: SYSTEM ANALYSIS, ALGORITHMS AND PROSPECTS FOR APPLICATION IN CONSTRUCTION DIAGNOSTICS
М. А. Filonova , S. N. Shirobokova188-2082026-07-07Abstract ▼The relevance of this study is determined by the growing interest in hyperspectral imaging as a tool for non-destructive testing of building materials and structures, as well as by the insufficient systematization of modern methods for processing such data. The aim of the work is to provide a systematic analysis of HSI processing algorithms, identify their advantages and limitations, and determine the prospects for their application in construction diagnostics. The study examines the specific features of hyperspectral data, including high dimensionality, noise, calibration errors, atmospheric distortions, and the shortage of labeled datasets. The evolution of approaches is shown: from classical machine learning methods and manual feature engineering to deep neural networks. Dimensionality reduction methods, kNN classifiers, Bayesian models, logistic regression, Random Forest, SVM, and MLP are analyzed, along with methods for incorporating spectral-spatial context. Special attention is paid to modern deep learning architectures: 1D, 2D, and 3D CNNs, RNNs, LSTM/GRU models, hybrid CNN–RNN models, transformers, and CNN–Transformer schemes. Transfer learning, semi-supervised learning, self-supervised learning, few-shot learning, meta-learning, and domain adaptation are considered separately as ways to overcome the limited availability of labeled data. The approaches are compared in terms of data requirements, computational complexity, robustness to noise, and their ability to account for spectral and spatial dependencies. It is shown that the most promising models for construction diagnostics are hybrid models that combine local convolutional features, the global context of attention mechanisms, and the possibility of fine-tuning on small specialized datasets. The paper summarizes the current state of the field and forms a basis for selecting methods for defect detection, moisture assessment, corrosion analysis, and evaluation of degradation in building structures. The conclusions formulated in the study can be used when designing experimental protocols and selecting architectures for further applied research in the field of building monitoring
-
A HYBRID APPROACH FOR DEEP LEARNING BASED FINGER VEIN BIOMETRICS TEMPLATE SECURITY
Shendre Shivam , Shubhangi Sapkal2020-10-11Abstract ▼We are living in the today’s society, where we have fairly-enough storage capacity and processing
power, the only issue is with security. As, the technologies are evolving with faster rate, we
are tend to grow the use of electronic devices rapidly in todays’ society, it started to flow or leakage
of personal information around/across, which then leads to breach of this information. Now,
personal or identical verification is key problem is being crucial. So whatever traditional methods
we have for providing authentication or security those have proven inadequate to be unreliable
and do not provide strong security. Biometric template protection is one of the most important
issues in securing today’s biometric system. We have many algorithms which don’t give adequate
solution for the same. So we tried to give a method which will reach to the expectations more satisfactorily
and certainly to the extent required. In this paper we have discussed a hybrid method for
finger vein biometric recognition based on deep learning approach using BDD and fuzzy commitment
schemes. The proposed hybrid method consists of four parts, namely Finger vein feature
extraction, BDD-based secure template generation, Fuzzy commitment scheme and ML based
finger vein recognition and decision making. Thus it has four module and each module works efficiently
and gives accurate results on all databases. -
CONTROL OF A MOBILE ROBOT ON BASE OF NEURAL NETWORK FOR THE PATH PLANNING IN UNMAPPED OBSTRUCTED ENVIRONMENT
А. К. Farhood99-1142022-01-31Abstract ▼In this work, a neural network of deep learning of a special structure is used. The neural
network allows a mobile robot to move without encountering obstacles in an unknown environment.
The main problems that the efforts of researchers in the field of neural network traffic planners
are aimed at solving are improving the performance of neural networks, optimizing their
structure and automating learning processes. The main result of this article is a new iterative algorithm
for developing a training set. At the first iteration, the initial training set is developed and
the initial training of the neural network is performed. In the following iterations, the neural network
trained at the previous stage is used as a filter for the following training sets. The filter selects
trajectories with collisions caused by neural network errors. During the learning process, the
number of convolutional and fully connected layers increases iteratively. Thus, the proposed algorithm
makes it possible to develop both a training set and a neural network architecture. Training
results are compared for filtered and unfiltered sets. The high efficiency of filtering has been confirmed,
as a result of which the distribution of examples in the training sample changes. The algorithm
can be used to develop a planning block for a mobile ground control system. The article
provides an example of training a neural network in a Matlab modeling environment. In the example,
five iterations of training were carried out, during which an accuracy of more than 90% was
achieved. This accuracy was obtained using the collected statistics on the movement of the mobile
robot in a randomly generated environment. The density of filling the environment with obstacles
was up to 40%, which corresponds to urban conditions. The comparison of neural network planners
trained using the proposed iterative procedure and with conventional training is carried out.
The comparison showed that the use of an iterative procedure increases the accuracy of planning
up to 12-15%. At the same time, the initial volume of the resulting sample is reduced several times
due to the applied filtering. -
MACHINE LEARNING AND DEEP LEARNING MODELS FOR ELECTRONIC INFORMATION SECURITY IN MOBILE NETWORKS
Aussi Rim Mohammed Hedhair, E.V. Zargaryan, Y.A. Zargaryan2022-08-09Abstract ▼Recent advances in wireless communication technologies have led to the creation of a huge
amount of data that is transmitted everywhere. Most of this information is part of an extensive and
publicly accessible network that connects various stationary and mobile devices around the world.
The capabilities of electronic devices are also increasing day by day, which leads to more data
generation and information exchange through networks. Similarly, with the increasing diversity
and complexity of mobile network structures, the frequency of security breaches in it has increased.
This hinders the introduction of intelligent mobile applications and services, as evidenced
by the wide variety of platforms that provide data storage, data computing and application services to end users. In such scenarios, it becomes necessary to protect data and check their use in
the network and applications, as well as check their incorrect use in order to protect private information.
According to this study, a security model based on artificial intelligence should ensure
the confidentiality, integrity and reliability of the system, its equipment and protocols that control
the network, regardless of its creation, in order to manage such a complex network as a mobile
one. The open difficulties that mobile networks still face, such as unauthorized network scanning,
fraudulent links, etc., have been thoroughly studied in this article. This article also discusses several
ML and DL technologies that can be used to create a secure environment, as well as many
cybersecurity threats. It is necessary to address the need to develop new approaches to ensure a
high level of electronic data security in mobile networks, since the possibilities for improving the
security of mobile networks are limitless. -
OPTIMIZATION OF PID PARAMETERS OF SERVO SYSTEMS USING A GENETIC ALGORITHM AND A NEURAL NETWORK CLASSIFIER
Ahmad Zoualfikar , Y.А. Kravchenko , А.М. Mansour237-2502025-10-01Abstract ▼Machine learning algorithms play a vital role in enhancing the performance of industrial systems, providing high precision and operational efficiency in real time. In servo motor control systems, these algorithms help reduce noise and vibration, improving efficiency and extending equipment lifespan. This article examines various types of noise that occur and their negative impact on industrial processes. The primary research objective is to optimize PID controller parameters in servo systems using a combined algorithm that combines neural networks and genetic algorithms. Unlike traditional methods such as genetic algorithms (GA) and particle swarm optimization (PSO), which suffer from slow convergence and risk of motor damage, the proposed solution is based on a control software platform. This platform ensures safe real-time interaction with the servo motor. A CAN Bus-based control system has been developed that enables developers to: read all servo motor parameters (speed, current, voltage, encoder position); modify PID coefficients with a single click, eliminating the need for manual tuning as in MOTO-MASTER. The implementation of the developed control system allowed the use of a trained neural classifier to constrain PID parameters within safe limits, reducing search space and accelerating the optimization process. Experimental results on SPH-S servo motors demonstrated significant reduction in noise and mechanical vibrations during real-time operation while maintaining stability across a wide speed range (0-1500 rpm).
-
IMPLICIT THREATS IDENTIFICATION BASED ON ANALYSIS OF USER ACTIVITY ON THE INTERNET SPACE
V. V. Bova , D. Y. Zaporozhets, Y.A. Kravchenko , E. V. Kuliev , V. V. Kureichik , N. A. Lyz2020-10-11Abstract ▼The article is devoted to the problem of identifying implicit information threats of a user's
search activity in the internet space based on an analysis of his activity in the course of this interaction.
The use of knowledge stored in the Internet space for the implementation of criminal intentions
poses a threat to the whole society. Identifying malicious intent in the users’ actions of the
global information network is not always a trivial task. The proven technologies for analyzing the
context of user interests fail in the case of cautious and competent actions of attackers who do not
explicitly demonstrate the goal they are pursuing. The paper analyzes the threats associated with
certain scenarios for the implementation of search procedures that manifest themselves in search
activities. Criteria of inefficient and effective search scenarios estimation are described. Among
the signs indicating the possibility of a threat, the following main ones are highlighted: avoiding
solving the problem in aimless navigation or attractive resources, superficial search, lack of
meaningful immersion in solving the search problem, and chaotic actions during the search.
To determine the presence of adverse signs, a system of indicators is built. The features of an effective
scenario for organizing a search in the Internet space are formulated, options for the presence
of implicit threats for a similar situation are described.An approach for identification the
described threats is presented taking into account the specified criteria for evaluating various
scenarios of user behavior in the global information space. A machine learning algorithm has
been developed to identify problem scenarios by comparing with key behavioral patterns. The
software implementation of the subsystem for identifying information threats has been created,
experimental studies have been conducted to confirm the effectiveness of the subsystem. Experimental
studies were carried out on the basis of processing open data from social networks, as well
as using analysis of user search activity in the university corporate information environment. -
INTELLIGENT METHOD OF KNOWLEDGE EXTRACTION BASED ON SENTIMENT ANALYSIS
E.M. Gerasimenko, V.V. Stetsenko2020-11-22Abstract ▼The paper explores the impact of age and gender in sentiment analysis, as this data can help
e-commerce retailers increase sales by targeting specific demographic groups. The data set used
was created by collecting book reviews. A questionnaire was created containing questions about
preferences in books, as well as age groups and gender information. The article analyzes segmented
data on the subject of moods depending on each age group and gender. Sentiment analysis
was performed using various machine learning (ML) approaches, including maximum entropy,
support vector method, convolutional neural network, and long short-term memory. This paper
investigates the impact of age and gender in sentiment analysis, because this data can help
e-commerce retailers to increase sales by targeting specific demographic groups, as well as increase
the satisfaction of the needs of people of different age and gender groups. The dataset used
is generated by collecting book reviews. A questionnaire was created containing questions about
preferences in books (user opinions of e-books, paperbacks, hardbacks, images and audiobooks),
as well as data on age group and gender. In addition, the questionnaire also contains information
on a positive or negative opinion regarding preferences, which served as the basis for reliability
for the classifiers. As a result, 900 questionnaires were received, which were divided into groups
according to gender and age. Each specific group of data was divided into training and test one.
Segmented data were analyzed for sentiment analysis depending on age group and gender.
The age group “over 50 years old” showed the best results in comparison with all other age
groups in all classifiers; data in the female group performed higher accuracy compared to data from
the groups without gender information. The high scores shown by these groups indicate that sentiment
analysis approaches are able to predict moods in these groups better than in others. Sentiment
analysis was performed using a variety of machine learning (ML) approaches, including maximum
entropy, support vector machines, convolutional neural networks, and long short term memory. -
ANALYSIS OF ENCRYPTED NETWORK TRAFFIC BASED ON ENTROPY CALCULATION AND APPLICATION OF NEURAL NETWORK CLASSIFIERS
V.A. Bukovshin, P.A. Chub, D.A. Korochentsev, L.V. Cherkesova, N.V. Boldyrikhin, O.A. Safaryan2021-02-13Abstract ▼Network traffic analysis allows you to solve many problems, such as: determining the pattern
of data transmission over the network, collecting statistics on the use of web applications,
monitoring and further researching network load, identifying potential malicious software and
network attacks, etc. 40% of Internet traffic belongs to unknown applications. This suggests that
for the area of network traffic analysis, the task of classifying applications has acquired particular
importance. Improvements in software in the field of network technologies have contributed to the
discovery of serious vulnerabilities in the implementation of some network protocols, namely TCP
and HTTP. By using network traffic analyzers, an attacker gained access to the contents of data
packets transmitted over the network. However, with the increasing qualifications of the information
community in the field of computer security, as well as with the development of network
technology standards, the analysis of network traffic has become noticeably more complicated.
The increased use of mathematical methods for protecting information, such as symmetric and
asymmetric cryptographic protocols, has led to the fact that most approaches to the analysis of
network traffic have lost their meaning and are no longer used. Therefore, the search for new
solutions to the problem of classifying network traffic, taking into account the possibility of its
encryption, is relevant. The article is devoted to the description of a new mixed approach to the
analysis of network traffic, based on the combined use of information theory and machine learning
algorithms. It also provides a comparative analysis of the proposed method with existing approaches
based on both information theory and machine learning. The aim of the research is to
develop an algorithm based on an intelligent approach to the analysis of network traffic. The proposed
algorithm is based on calculating entropy and using neural network classifiers. Research
objectives include: theoretical substantiation of the proposed approach in the field of information
theory, as well as machine learning algorithms; carrying out a structural description of the implemented
algorithms for calculating entropy and classifying applications that generate encrypted
traffic; comparative analysis of the proposed algorithm with existing approaches to the analysis of
encrypted network traffic. The result of the research is a new algorithm that allows classifying
various types of encrypted traffic with a high degree of reliability. -
OBJECT DETECTION ALGORITHM FOR OPTOELECTRONIC SYSTEMS WITH ONLINE LEARNING
V.A. Tupikov, V.A. Pavlova, V.A. Bondarenko, M.V. Sozinova, P.A. Gessen2021-04-04Abstract ▼In order to create a new algorithm for automatic detection of objects with real-time training, a
study of the world scientific groundwork in the field of general-purpose automatic tracking with the
ability to recognize a tracked object with the potential for application in embedded computing systems
of optoelectronic systems of promising robotic complexes was carried out. Based on the conducted
research, methods and approaches were selected and tested that allow, with the greatest accuracy,
while maintaining high computational efficiency, to provide training of classifiers on the fly
(online learning) without a priori knowledge of the type of tracking object and to ensure the subsequent
detection of the original object in the event of its short-term loss. Such methods include a histogram
of oriented gradients – a descriptor of key features based on the analysis of the distribution of
the brightness gradients of the object image. Its use allows you to reduce the amount of information
used without losing key data about the object and to increase the speed of image processing. The
article substantiates the choice of one of the real-time classification algorithms that allows solving
the problem of binary classification – the support vector machine. Due to the high speed of data processing
and the need for a small amount of initial training data to construct a separating hyperplane,
on the basis of which the classification of objects is done, this method is chosen as the most suitable
for solving the problem. For online training, a modification of the support vector machine method
was chosen, which implements stochastic gradient descent at each step of the algorithm – Pegasos.
The authors of the study carried out the development and semi-natural modeling of the selected algorithm,
evaluated the effectiveness of its work in the tasks of detecting an object of interest in real time
with preliminary online training in the process of tracking the object. The developed algorithm has
shown high efficiency in solving the problem and is planned to be implemented as part of a special
software for optoelectronic systems of advanced robotic systems. In the conclusion, proposals are
presented to further improve the accuracy and probability of the object detection by the developed
algorithm, as well as for improving its performance by optimizing calculations. -
MACHINE LEARNING MODEL OF SWARM EVASION FROM THE INFLUENCE OF ANTAGONISTIC ENVIRONMENT
V. К. Abrosimov, G.А. Dolgov, Е. S. Mikhailova6-192025-04-27Abstract ▼One of the priority areas of group control theory for the near future is swarm control of groups of
small unmanned aerial vehicles - micro-, mini- and nano-classes, performing a collective task under enemy
influence. Here, two antagonistic strategies collide - minimization of losses from the point of view of
the attacking swarm and maximization of such losses from the point of view of the defense system. Research objective: development of an approach to solving a practical problem - penetration of a swarm of
unmanned aerial vehicles into an object protected by a defense system. The objectives of the study were to
analyze the characteristics of the factors influencing the processes of detection, tracking, recognition of
swarm intentions by the defense system and the development of a machine learning model for creating
spatio-temporal formations that minimize the number of swarm elements affected by the defense system.
The main parameters of the defense system are the detection range and duration of swarm recognition, the
time to make a decision on the actions of the swarm, the size of the zone of destruction of defense means.
The method of machine learning on convolutional neural networks with reinforcement was chosen as the
research method. The counteraction effect against the defense system is created due to the swarm's dynamics;
it can actively maneuver, creating spatio-temporal maneuvers during the mission. To simulate the
"Swarm vs. Defense System" situation, a swarm agent (a neural network with a transformer architecture
that initiates swarm formations) and a defense system agent are introduced that recognizes the swarm and
attacks it, creating a zone of destruction in the conventional center of mass of the swarm. The swarm is
guided by a stochastic rule, asking the defense system (environment) to react to its maneuver. The environment
responds by attacking the swarm, creating a damaging factor at the point where the swarm or the
main part of the swarm is expected to be. The reward of the swarm strategy is the number of undestroyed
objects under the conditions of constraints; for the defense system, this "reward" acts as a "punishment".
An interesting phenomenon was established in the process of machine learning: each swarm element,
remaining within a given space and implementing the biological principles of swarm control without a
Leader, independently evades the area of destruction, which together creates a random spatio-temporal
formation for defense means with minimal losses of swarm elements. Thus, using the method of machine
learning with reinforcement, a model was created that allows varying the behavior of the swarm and synthesizing
spatio-temporal formations that complicate detection, tracking, recognition of intentions and
decision-making on the impact of the defense system on a swarm of attacking small unmanned aerial vehicles,
as well as significantly reducing their losses. -
DEVELOPMENT OF INTELLIGENT MOBILE APPLICATIONS
Т. А. Kramarenko, E. V. Feshina , T. V. Lukyanenko2022-06-03Abstract ▼The article presents the development results of a module for the retail network mobile application
modernization. A feature of the presented mobile application module is the display of personalized
messages with advertising and promotions of the retail network. A mathematical model
of machine learning is used to collect and analyze data in a mobile application. The process ofchoosing a mathematical model, the operation algorithm and the model training stages on training
data are described in detail. The quality of the classifier's work was evaluated on a test and training
sample. Test sample objects classification and the real value of the class comparison with the resulting
classification were performed. The authors in the article presented the main stages of the algorithms
development for processing statistical data from customer receipts. The program codes for the
receipt analysis module implementation and display the mobile application personalized advertising
are presented. To implement the database as a tool, the authors used the relational data management
system MS SQL Server. The modules of the mobile application are developed in the Android Studio
environment for the Android operating system family. The authors presented the algorithm main
stages and testing the implemented modules operability in the paper. Based on the data on purchases
made by the buyer, information about preferred products is collected based on the fixation of product
groups and product items from the receipt. The loyalty card of the retail network is linked to the mobile
application, and receipts for purchases are linked to loyalty cards, in turn. Previously, the application
displayed ads for all products participating in promotions. The actual task is to display personalized
advertising, which has proven its effectiveness. The mobile application is distributed for
free through the Play Market and is designed for smartphones running the Android OS line.
The purpose of the development is to display in the application on the buyer's device first advertising
frequently purchased goods, and then the rest of the promotional goods. The mobile application has
passed load testing in real use by customers conditions of the retail network. -
FAILURE PREDICTION USING FACTOR ANALYSIS METHODS
Е.S. Podoplelova213-2232025-07-24Abstract ▼This article discusses the application of a risk assessment method based on the combination of the FMEA (failure mode and effect analysis) methodology and the MCDM (Multiple Criteria Decision Making) methods. This approach allows taking into account both expert knowledge and historical data on the operation of the equipment. MCDM methods process the assessment more flexibly in comparison with the standard method of calculating the priority number of risks (PRN), which helps to better assess the risks by three criteria: the probability of occurrence, the complexity of detection and the severity of the consequences. One of the criteria can be obtained not only through an expert assessment, but also on the basis of data recording the operation of the equipment. This approach was tested using the example of synthetic open-source data on the operating modes of production equipment. The task was to predict both the failure itself and its type, as well as to identify the factors that have the greatest impact on the failure. For this purpose, data preprocessing was carried out, during which it was necessary to eliminate the imbalance of classes. There are several approaches to solving this problem, aimed at reducing the dominant class or generating instances of poorly represented classes. In this example, random reduction of the number of records without errors was used. Then, AdaBoost, Random Forest and LinearSVC were compared as classification algorithms. Since multi-class classification was required, it was decided to use the one-vs-the-rest strategy. As a result, it was possible to achieve 86% forecasting accuracy by F-measure using the AdaBoost and Random Forest algorithms. LinearSVC turned out to be ineffective. Thus, the resulting forecasting model recognizes different types of errors, but there is room for improvement, which requires a larger sample, including more examples with different types of failure. Based on this, this approach as an alternative to expert assessment is promising, improving objectivity, and also making it possible to foresee risks and prevent a real failure or risk-related incident.
-
RESEARCH OF MACHINE LEARNING METHODS FOR DETECTING SPOOFING ATTACKS IN DECENTRALIZED NETWORKS
М.А. Lapina , R.А. Dymuha , N.N. Kucherov , Е.S. Basan16-312025-07-24Abstract ▼Unmanned aerial vehicles are appearing more and more in our lives and are used for various purposes such as cargo delivery, monitoring, household management, exploration and entertainment. But along with their growing popularity, the number of people who intentionally want to interfere with the operation of UAVs and use them for their own interests and purposes is also increasing. They use various types of attacks to eliminate or intercept the drone by any means. Spoofing attacks are one of the most common and dangerous types of attacks, as they allow attackers to act unnoticed, faking the identifiers of autonomous aircraft or operators, posing as legitimate participants in the system. The purpose of such attacks may be to intercept control, steal data, sabotage, or use UAVs to perform malicious actions such as espionage, damage, or malfunction operations. But every year it becomes more difficult to prevent attacks, as they are difficult to detect and can lead to serious consequences, which is why such a solution as detecting spoofing attacks on an unmanned vehicle using machine learning was invented. The article discusses spoofing attacks on UAVs, analyzes spoofing on autonomous aircraft, and studies machine learning methods for detecting spoofing attacks based on a dataset using the Knime platform. The results of the study demonstrate that the method of detecting attacks using machine learning based on the ensemble method, the Tree Ensemble Learner and Random Forest Learner models, which showed results of 97.110% and 97.039%, respectively, is the best among other methods, which will improve the security of unmanned aerial vehicles, reduce the burden on operators and increase the reliability of the system as a whole. In the future, the proposed approach can be expanded to detect other types of cyberattacks, which will make it a universal method of protection against intruders
-
TEXT VECTORIZATION USING DATA MINING METHODS
Ali Mahmoud Mansour , Juman Hussain Mohammad, Y. A. Kravchenko2021-07-18Abstract ▼In the text mining tasks, textual representation should be not only efficient but also interpretable,
as this enables an understanding of the operational logic underlying the data mining
models. Traditional text vectorization methods such as TF-IDF and bag-of-words are effective and
characterized by intuitive interpretability, but suffer from the «curse of dimensionality», and they
are unable to capture the meanings of words. On the other hand, modern distributed methods effectively
capture the hidden semantics, but they are computationally intensive, time-consuming,
and uninterpretable. This article proposes a new text vectorization method called Bag of weighted
Concepts BoWC that presents a document according to the concepts’ information it contains. The
proposed method creates concepts by clustering word vectors (i.e. word embedding) then uses the
frequencies of these concept clusters to represent document vectors. To enrich the resulted document
representation, a new modified weighting function is proposed for weighting concepts based
on statistics extracted from word embedding information. The generated vectors are characterized
by interpretability, low dimensionality, high accuracy, and low computational costs when used in
data mining tasks. The proposed method has been tested on five different benchmark datasets in
two data mining tasks; document clustering and classification, and compared with several baselines,
including Bag-of-words, TF-IDF, Averaged GloVe, Bag-of-Concepts, and VLAC. The results
indicate that BoWC outperforms most baselines and gives 7 % better accuracy on average -
CONCEPTUAL MODEL OF FACTORS INFLUENCING THE EFFICIENCY OF GAS PREPARATION AND SEPARATION PROCESS
А. V. Martirosyan , D. V. Romashin241-2492026-09-10Abstract ▼The paper presents the concept of adaptive control in natural gas separation. The control systems used in practice are usually based on fixed control algorithms and do not consider dynamic changes in physical, technical and operational parameters, which leads to a decrease in separation quality and an increase in energy consumption. The article describes the approach of combining system analysis, modeling and the Pareto method. This paper presents a concept for adaptive control in natural gas separation. Control systems used in practice are typically based on fixed control algorithms and do not account for dynamic changes in physical, technical, and operational parameters, which leads to reduced separation quality and increased energy costs. This paper describes an approach combining systems analysis and the Pareto method. The aim of this paper is to develop a concept for adaptive control of the natural gas treatment process based on systems analysis. Recent studies demonstrate that data-driven methods enable more precise parameter adjustment, better responsiveness to raw-gas fluctuations and improved impurity removal efficiency. To achieve this goal, this paper addresses the challenges of identifying and classifying factors affecting the quality and efficiency of gas separation, as well as integrating their relationships within a unified conceptual control model. Particular attention is paid to the impact of precise control of key parameters, such as pressure, temperature, and flow rate, on the efficiency of these processes. An analysis of recent research demonstrates the growing use of neural networks and machine learning models in gas purification for predictive control, anomaly detection, and optimization of operating parameters. A comparative evaluation of classic PID controllers, fuzzy, adaptive, and neural control methods confirms the advantages of intelligent control in terms of stability, adaptability, and energy efficiency. The main result of this study is the substantiation of key factors determining separation efficiency, among which pressure, temperature, and gas flow rate have the greatest impact. The resulting model forms a methodological basis for the development of intelligent and adaptive control systems for gas purification processes.
-
MODERN APPROACHES TO SOLVING THE 3D BIN PACKING PROBLEM
М.М. Sorokin , L. А. Gladkov , N. V. Gladkova131-1502026-09-10Abstract ▼The article is devoted to the consideration of current trends and approaches to solving the urgent optimization problem of three-dimensional bin packing problem. The importance of building effective methods for solving this problem is due to the rapid growth of e-commerce, where achieving even incremental improvements in container filling density can lead to significant reductions in freight transportation and storage costs. The article provides an analysis of various types of problems and suggests a classification of bin packing problems according to various criteria, including: offline and online packing, by dimension, by type and quantity of containers and cargo. The formulation of the classical optimization knapsack problem is given and various options for constraints due to the specifics of the tasks being solved are considered. A brief overview of the main approaches to solving the problem is given. The analysis and generalization of the characteristic features of the application of metaheuristic approaches based on the use of evolutionary and bioinspired algorithms and machine learning methods is carried out, their advantages and disadvantages are noted. Due to the complexity of the problem under consideration, it is proposed to actively use known and develop new modifications of metaheuristic algorithms that make it possible to find quasi-optimal solutions in polynomial time. The analysis of known machine learning methods and bioinspired algorithms is given, the principles of their operation are described, their main features, advantages and disadvantages are highlighted, and the prospects for their development and application to solve NP-complete combinatorial optimization problems are noted. A generalized principle of operation of metaheuristic algorithms is given. A comparative analysis of the application of various optimization methods has shown the effectiveness of using metaheuristic methods to solve the problem of three-dimensional packaging.
-
IMPROVING THE QUALITY OF MULTI-MODAL DATA BASED ON THE HYBRIDIZATION OF PREPROCESSING AND MULTISENSORY FUSION METHODS
А. А. Aleksandrov , М.А. Butakova21-382026-09-10Abstract ▼The article explores the methods of data preprocessing and merging of multi-modal data. The data collected from multi-sensor devices contains noise, anomalies, and includes sensor failures. The use of raw data inevitably leads to false patterns when using machine learning models in which this data is used. To improve the quality of data, it is necessary to use special methods for preliminary processing and merging. The aim of the work is to create a hybrid preprocessing and multisensory data fusion controller operating at the peripheral computing level. To achieve this, the tasks of removing the consequences of technical failures and noise from the data, time synchronization, scaling of features and combining duplicate sources, considering their dynamic weight, have been solved. The following methods were applied: inter-quartile scale to detect outliers, linear interpolation to replace them, high-pass filter to suppress noise, and weighed multisensory fusion with dynamic weight calculation. In addition, methods for processing visual, sensory, and acoustic data within a hybrid pipeline are presented that can be adapted for implementation at the peripheral computing level. The experimental verification of the developed pipeline was carried out on a simulated set of climatic data with artificially introduced noises and anomalies.
The proposed approach has demonstrated high efficiency of data recovery. For air temperature, the coefficient of determination (R2) increased from 0.9807 to 0.9964 with a slight change in the mean absolute percentage error (MAPE) from 10.11% to 10.48%. For relative humidity, the R2 metric increased from 0.9248 to 0.9675, and the MAPE error decreased from 2.87% to 2.57%. The most significant improvement has been achieved for atmospheric pressure. The R2 metric increased from a negative value of -0.124 to 0.974, and the MAPE error decreased from 0.12% to 0.06%. The results confirm that the proposed pipeline reduces the error and improves the quality of multisensory data, minimizes the amount of data being transferred. The practical value lies in creating a continuous cycle of local filtering and merging of data without accumulation of errors. -
ALTERNATIVE APPROACHES TO NLP MODEL SCALE-UP: AN ANALYSIS OF APPROACHES TO OPTIMIZING DATA AND COMPUTATION VOLUME WHEN TRAINING LARGE-SCALE LANGUAGE MODELS
К.I. Ralko , N. Е. Sergeev152-1722026-07-07Abstract ▼This paper focuses on overcoming the systemic limitations of the large-scale language model (LLM) scaling paradigm, which are related to data exhaustion and exponential growth in computational costs. This enables the development of more efficient approaches to building NLP models without sacrificing their performance. The goal of this study is to compare the performance of a standard transformer architecture (nanoGPT) and a model using semantic embeddings (nanoSonar) for language modeling tasks under resource constraints. Working with conceptual embeddings allows us to identify deeper linguistic patterns and reduce the amount of required training data, significantly improving modeling efficiency. The study utilized the TinyStories dataset, which includes short narratives with a clear structure. Before implementing the models, the data was preprocessed: for nanoGPT, tokenization was performed using the BPE method, and for nanoSonar, text was converted into semantic embeddings using a pretrained Sonar model. The models were evaluated using the loss and perplexity metrics. The results showed that the nanoSonar model provides significantly lower perplexity (6.609 versus 39.151 for nanoGPT) and demonstrates more robust training dynamics at later stages. This paper presents an analysis of modern approaches to scaling optimization (MoE, distillation, PEFT) and promising architectures (LRM, SSM, RWKV), and provides practical recommendations for applying models operating in the space of semantic embeddings to domain-specific problems and systems with limited computational resources. The results of this study can be useful in developing efficient language models that combine high generation quality with a cost-effective architecture.
-
FORECASTING STUDENT MOVEMENT USING MACHINE LEARNING AND TIME SERIES ANALYSIS
Mirziyod Adkham ugli Radjapov , К.D. Chemukhin , L. E. Petrosyan134-1512026-07-07Abstract ▼Managing student mobility amid demographic fluctuations and the digitalization of higher education is becoming a key factor in university sustainability, affecting both financial performance and the quality of the educational process. The increasing complexity of processes such as admissions, withdrawals, academic leaves of absence, transfers, and reinstatements requires a shift from expert assessments to formalized models and predictive analytics based on the processing of large datasets. The aim of this study is to develop and evaluate the effectiveness of a model for forecasting student population dynamics based on machine learning algorithms and using time series analysis. Aggregated statistical data on student mobility at Russian and Chinese universities for the period 2013–2024 were used as the empirical basis, which allowed for consideration of both the structural features of national higher education systems and long-term trends and anomalous events (including the impact of the COVID-19 pandemic). The methodological framework includes a dynamic student cohort balance model in the form of a system of recurrent equations describing transitions between academic years and enrollment statuses, and an additive Prophet model used for independent forecasting of key flows (admissions, transfers, withdrawals, academic leaves of absence, reinstatements) as separate time series. The software implementation is based on the FastAPI–React stack, utilizing the SQLAlchemy ORM layer and mechanisms for caching the results of predictive calculations, which ensures high performance when processing queries. Experimental results on real data demonstrate the robustness of the developed model to nonlinear changes in the input series and confirm the feasibility of integrating machine learning into the student movement management system. The practical significance of this work lies in the creation of an information and analytical system that provides automated monitoring and forecasting of student movement trajectories between courses and statuses, enabling universities to transition from reactive to proactive planning of admissions campaigns, classroom allocation, and the distribution of personnel and infrastructure resources.
-
SIGNAL DEMODULATION USING CLASSICAL MACHINE LEARNING ALGORITHMS FOR THE WATTERSON MODEL
А.S. Kirillov97-1052026-07-07Abstract ▼This paper investigates the application of classical machine learning algorithms for signal demodulation in a high-frequency communication channel described by the Watterson model. The relevance of this task stems from the need to improve noise immunity during broadband data transmission under conditions of multipath propagation and Doppler distortions. The classifiers used include Random Forest, Decision Tree, and k-nearest neighbors (KNN). The models are trained on synthesized data generated in the MATLAB Simulink environment, with varying path delays (0–4 ms) and Doppler shifts (0–10 Hz).
The input features comprise the coordinates of signal constellations as well as the history of received symbols, since intersymbol interference significantly affects the position of the current point. The experiments revealed a critical feature: demodulation algorithms exhibit high sensitivity to variations in Doppler shift. A deviation of just 0.1 Hz from the training set parameters alters the structure of the signal constellation, increasing the bit error rate (BER) to 0.5. When the Doppler parameters align with the training grid, the Random Forest algorithm demonstrates the best performance, achieving a BER <0.01 on the full dataset. To address this sensitivity, a local training method with a reduced Doppler shift sampling step (down to 0.1 Hz) is proposed, enabling a BER ≤0.01 for Random Forest. Due to the exponential growth of the training dataset size (up to 1000 GB), data reduction techniques were developed: excluding time parameters and sampling data based on possible combinations of preceding bits. This reduced the data volume by a factor of 31.5 while maintaining a BER <10% for Random Forest with a training step of 0.2 Hz. The study concludes that the Random Forest classifier is suitable for demodulation in HF channels and highlights the necessity of adapting the training step to the required Doppler shift accuracy. -
IMAGE MATCHING SYSTEM WITH USING INTUITIONISTIC FUZZY SETS
К.I. Morev286-2982026-04-29Abstract ▼This paper presents a fully learnable system for solving the problem of matching two images. All the main elements of the system are trainable, i.e. their final form corresponds to the target dataset on which the training was carried out. The fact that the system is trainable, the methods used in training and the architecture of the system allow using the system to solve a large number of various computer vision problems. The system consists of a convolutional neural network that serves both to extract key points and their descriptors, as well as a trainable matcher of the extracted key points based on their description and mutual arrangement in the observed scene. The used convolutional neural network processes full-size images and calculates both the location of interest points with pixel accuracy and the descriptors associated with them in a single forward pass. Matching key points is a separate step and is performed after the forward pass of the neural network. In the process of training the model for calculating the positions of key points and their descriptors, a method for forming a training sample is used, called homographic adaptation - an approach that helps to increase the repeatability and accuracy of detecting key points. The process of training the feature point detection model consists of obtaining new weights in the process of additional training of the base detector, which represents the initialization weights of the model. The final feature point detection model, trained on the universal MS-COCO image set using homographic adaptation, repeatedly outperforms the original base detector in terms of the number, reliability and repeatability of feature points, and also outperforms any other traditional corner detector based on classical approaches
-
A UNIVERSAL MODEL OF ADAPTIVE MANAGEMENT OF CLOSED AGRICULTURAL PRODUCTION USING AI TECHNOLOGIES
А.А. Kochkarov , А.К. Kulikov , V.М. Matsakova2026-04-29Abstract ▼The relevance of this research stems from the contradiction between the need to increase food production in an urbanized environment and the fragmentation of existing high-tech solutions (hydroponics, aeroponics, IoT), which are being implemented in isolation, without a unified management methodology. The lack of unified approaches to data collection and adaptive control of environmental parameters limits the scalability of vertical farms. The goal of this research is to develop and theoretically substantiate the architecture of a universal adaptive management model for closed-loop agricultural production systems, integrating various cultivation methods based on machine learning algorithms. The methodology is based on a systematic analysis of scientific publications and experimental data on the use of embedded devices and machine learning algorithms in hydroponic, aeroponic, and soil-based vertical greenhouses. Based on this data synthesis, parametric matrices were constructed to standardize technological processes. The main results include the development of a structural diagram of a universal model that enables the integration of disparate systems into a single platform with the ability to continuously monitor and perform predictive analytics. The minimum required sensor set is substantiated: pH, EC, temperature, humidity, CO₂, PAR, pressure, and nutrient solution flow. The proposed architecture enables dynamic switching between hydroponic, aeroponic, and indoor modes within a single phytotron. The conclusions and significance of this work lie in creating a foundation for designing scalable vertical farms with predictable profitability and resource efficiency indicators, while enabling continuous further training of AI algorithms for predictive microclimate management and early plant disease detection in urban environments
-
RESEARCH OF MACHINE LEARNING METHODS FOR DETECTING FRAUDULENT WEBSITES
М.А. Lapina , D. А. Lukyanov , V.G. Lapin , N.N. Kucherov250-2622025-10-01Abstract ▼Every year our lives become more and more connected with large volumes of data that need to be analyzed. As the volume of information increases, its analysis becomes a more voluminous and complex task.
In this situation, the problem of finding a tool that will help companies and institutions in collecting, analyzing and forecasting data arises. Machine learning is an area of artificial intelligence that finds patterns in a database and, based on them, tries to predict the result. Another area of application of machine learning is the detection of fraudulent sites. Currently, with the development of information technology, digital crimes have become a serious threat to confidential information and user data. Artificial intelligence is able to analyze site parameters and determine the presence of threats to information. The study is aimed at systematizing knowledge about phishing attacks and studying machine learning methods for detecting fraudulent sites. During the study, machine learning methods for detecting phishing sites were developed, schemes were built that allow machine learning models to correctly transform data for feeding them to models. The analysis of the data provided in the dataset made it possible to correctly transform the data for the correct operation of the models, which will avoid errors. The problem of retraining machine learning models was solved. A detailed study of the dataset made it possible to filter out data that could cause errors in the model and reduce the quality of artificial learning forecasting. As a result of the work, the developed methods for searching for phishing attacks using machine learning models were tested on test data, based on the results obtained, graphs of changes in the accuracy of detecting illegitimate sites from changing the model settings were constructed. An analysis of the study was carried out and the results of the work were summarized. -
METHODOLOGICAL SUPPORT FOR ASSESSING THE AVAILABILITY OF GOODS IN DISTRIBUTED STORAGE BASED ON COMPUTER VISION METHODS
А.R. Nedvigin , R.М. Sinetsky2026-02-27Abstract ▼This paper presents a formalization of the problem of automated monitoring of product availability on retail shelves and compliance with the prescribed planogram, leveraging computer vision and machine learning techniques. The aim of this research is to develop algorithmic solutions for the automatic assessment of product availability in distributed retail environments using computer vision methods, thereby addressing the challenge of maintaining optimal and necessary product assortments through continuous shelf monitoring and supporting data-driven managerial decision-making. A technological pipeline for visual data processing is proposed, comprising the stages of image normalization, segmentation, object localization, and classification, implemented with convolutional neural networks—specifically YOLO and U-Net architectures. An integrated product availability metric is introduced, which jointly accounts for physical, visual, and informational dimensions of availability. An optimization problem aimed at improving overall availability is formulated, and an adaptive neural network fine-tuning mechanism is implemented to enhance the accuracy of image recognition and segmentation, as well as the quality of analytical recommendations. Furthermore, an availability-improvement algorithm is proposed for a decision support system, based on the construction of an optimized merchandiser routing plan that prioritizes products and minimizes time expenditures. This routing problem is reduced to a generalized Traveling Salesman Problem (TSP) with priority-based weights. Methods for evaluating and enhancing product availability are proposed and described in detail. Based on the developed approaches and algorithms, a software system for monitoring and improving product availability has been implemented. Experimental results confirm the effectiveness of the proposed solutions: the average recognition accuracy reached 95.8%, and the integrated availability score achieved A = 0.93. The practical significance of this work lies in establishing an algorithmic foundation for intelligent shelf-monitoring systems that enable more efficient management of retail operations and inventory processes
-
ALGORITHM FOR SIGN LANGUAGE RECOGNITION USING CONVOLUTIONAL NEURAL NETWORK
V.Е. Bondareva , Т.S. Chernomorova , А.V. Krivtsun , Abdulkarem Abeer2026-02-27Abstract ▼This paper addresses the problem of automatic recognition of Russian Sign Language (RSL) using computer vision and deep learning methods. The relevance of the study is driven by a steady increase in the number of people with hearing impairments: according to the World Health Organization, there are currently about 70 million deaf and hard-of-hearing individuals worldwide, and this number is projected to reach 630 million by 2035. The development of effective gesture recognition algorithms is an important direction for creating contactless human–machine interaction systems aimed at improving accessibility of information technologies and enhancing the quality of life for people with hearing disabilities. The aim of the study is to develop and experimentally validate an algorithm for real-time recognition of Russian Sign Language alphabet gestures in a video stream using a convolutional neural network. A specialized dataset was created, consisting of 430 images of hand gestures corresponding to the letters of the RSL alphabet, captured from different angles and under varying lighting conditions. The model was implemented using TensorFlow and Keras libraries, while integration with the video stream was performed using OpenCV and a marker-based hand tracking system. As a result of training and testing, the proposed model achieved a recognition accuracy of 99% on the test dataset. A comparative analysis with classical machine learning methods demonstrated the superiority of the convolutional neural network in terms of classification accuracy and robustness to external noise. The obtained results confirm the effectiveness of the proposed approach and its applicability for real-time systems intended for communication, educational, and rehabilitation applications, as well as for the development of advanced human–machine interaction interfaces.
