Search

Search Results

• METAHEURISTICS BASED ON THE BEHAVIOR OF A COLONY OF WHITE MOLES

  Y.V. Danilchenko, V. I. Danilchenko, V. М. Kureichik

  132-140

  2021-08-12

  Abstract ▼

  Optimization algorithms inspired by the natural world have turned into powerful tools for solv-ing complex problems. However, they still have some disadvantages that require the study of new and more advanced optimization algorithms. In this regard, when solving NP complete problems, there is a need to develop new methods for solving this class of problems. One of these methods can be metaheuristics based on the behavior of a colony of white moles. This paper proposes a new metaheuristic algorithm called the blind white moles algorithm. This algorithm was developed based on the social behavior of blind moles in search of food and protecting the colony from intruders. The proposed solution will be able to overcome many disadvantages of conventional optimization algo-rithms, including falling into the trap of local minima or a low convergence rate. The purpose of this work is to develop an algorithm for optimizing a complex objective function. The scientific novelty lies in the development of a genetic algorithm based on the behavior of a colony of white moles for solving NP complete problems. The problem statement in this paper is as follows: to optimize the search for solutions to complex functions by applying an algorithm based on the behavior of a colony of white moles. The practical value of the work lies in the creation of a new search architecture that allows using the developed algorithm for the effective solution of NP complete problems, as well as conducting a comparative analysis with existing analogues. The fundamental difference from the known approaches is in the application of a new bioinspired search structure based on the behavior of a colony of white moles, which will allow to exclude falling into a local minimum or a low conver-gence rate. The presented results of the computational experiment showed the advantages of the pro-posed multidimensional approach to solving the problems of placing VLSI elements in comparison with existing analogues. Thus, the problem of creating methods, algorithms and software for solving NP complete problems is currently of particular relevance

• MULTILEVEL APPROACH FOR HIGH DIMENSIONAL 3D PACKING PROBLEM

  V. V. Kureichik, А. Е. Glushchenko

  2020-07-20

  Abstract ▼

  The article considers one of the important combinatorial optimization problems, the problem
  of 3D packing of different elements in a fixed volume. It belongs to the class of NP-complex and difficult
  optimization problems. The paper presents and describes the formulation of the 3D packing
  problem, introduces a combined objective function that takes into account all the restrictions. Due to
  the complexity of this task, a multilevel approach is proposed. It is consisting in dividing the 3D packing
  problem into 3 subtasks and solving each subtask in a strict order. Moreover, for each of the
  subtasks a unique set of objects is defined that are not repeated in the remaining subtasks. To implement
  a multi-level approach, the authors developed a combined bio-inspired algorithm based onevolutionary and genetic search. This approach can significantly reduce the time to obtain the result,
  partially solve the problem of preliminary convergence of the algorithms and obtain sets of quasioptimal
  solutions in polynomial time. A software package was developed and computer-based algorithms
  for automated 3D packaging based on a combined bio-inspired search were implemented.
  A computational experiment was conducted on test examples (benchmarks). The packaging quality
  obtained on the basis of the developed combined bio-inspired algorithm is on average 5 % higher
  than the packaging results obtained using known algorithms, and the solution time is less than 5 % to
  20 %, which indicates the effectiveness of the proposed approach. The series of tests and experiments
  carried out made it possible to refine the theoretical estimates of the time complexity of the packaging
  algorithms. In the best case the time complexity of the O (n2) algorithms; in the worst, O (n3).

• DEVELOPMENT OF BIOHEURISTICS FOR CREATING AN INTELLECTUAL SUBSYSTEM FOR MAKING EFFECTIVE DECISIONS OF NP-HARD AND NP-DIFFICULT COMBINATORY-LOGICAL PROBLEMS ON GRAPHS

  D. V. Zaruba , E. V. Kuliev , D.Y. Zaporozhets , M. M. Semenova

  2021-11-14

  Abstract ▼

  The article is devoted to the solution of new topical problems that have arisen in the conditions
  of the modern development of information and nanometer technologies in the field of design,
  as well as the development of new innovative methods that provide effective solutions in polynomial
  time. The article deals with the problem of solving NP-hard problems. The description of the
  procedure for measuring the complexity of the problem is presented the features of NP-hard and
  NP-difficult combinatorial logic problems are described. The main differences between the tasks
  are presented, as well as the problems that one has to face when solving this type of task. The general
  decision-making scheme is presented, consisting of the problem formulation; decisionmaking;
  signal in automatic systems and feedback. At the second stage (formation and selection of
  solutions), the solution is based on a bioinspired algorithm for finding solutions to the traveling
  salesman problem. To solve this problem, a modified bioinspired algorithm based on the behaviorof an ant colony was developed. Unlike other optimization methods, metaheuristic algorithms can
  find global optimal solutions for problems where there are many local solutions due to their random
  nature. These reasons have led to the widespread use of such algorithms in solving various
  optimization problems. Bioinspired algorithms are becoming a new revolution in the field of solving
  optimization problems. The statement of the traveling salesman problem is presented, as well
  as the solution of the problem on the basis of the ant algorithm. Algorithms such as genetic algorithms
  and PSO can be very useful, but they still have some disadvantages in solving multimodal
  optimization problems. These algorithms can find optimal solutions regardless of the physical
  nature of the problem. In the framework of experimental studies, the analysis of the work of
  bioinspired algorithms was carried out: the algorithm of a flock of bats, the bacterial algorithm
  and the ant algorithm.

• SOLUTIONS’ ENCODING IN EVOLUTIONARY METHODS FOR INSTRUMENTAL DESIGN PLATFORM

  E.V. Kuliev, А. А. Lezhebokov, М. М. Semenova, V.A. Semenov

  2020-07-20

  Abstract ▼

  The article considers current issues and analyzes the problems of three-dimensional integration
  and three-dimensional modeling that arise at the design stage during the solution of the
  problem of optimal planning of components of large and extra-large integrated circuits and case
  devices of electronic computing equipment. The main advantages of applying the principles of
  three-dimensional integration are presented and described in sufficient detail, which allow efficiently
  organizing the production of personalized electronics, optimally planning the configuration
  of large and ultra-large integrated circuits, taking into account thermal and energy characteristics.
  In the course of research, the authors developed an approach to encoding decisions based on
  an intelligent mechanism, which is characterized by the presence of built-in means of control of
  acceptable decisions. One of such tools that have experimentally proven their effectiveness is the
  built-in mechanism of “deadly mutations”, which takes into account the status of genes and predetermined
  restrictions on the final configuration of the housing of the designed device. A series of
  general approaches and specific algorithms for solving the planning problem based on the results
  of research by the author's team and modern approaches to solving NP-complete problems are
  proposed. The most important practically significant result of the research of the indicated problem
  is the developed software and instrumental design platform in the modern cross-platform Java
  programming language. The selected development technology allows you to use all the main advantages
  of modern multi-core and multi-processor architectures, to use software multi-threading
  to implement parallel schemes for solving combinatorial problems. The software and tool platform
  has a user-friendly interface, which allows you to effectively manage the process of solving the
  problem of planning the components of large and ultra-large integrated circuits of threedimensional
  integration by visualizing key performance indicators of algorithms on graphs and in
  text statistics blocks. The developed application software made it possible to carry out a series of
  computational experiments based on random data sets, as well as on open-data boron benchmarks
  for such tasks. The results of experimental studies have confirmed the theoretical estimates of the
  time complexity and effectiveness of the proposed approaches and algorithms, including the genetic
  algorithm, which uses the new decision coding mechanism proposed in the work.

• HEURISTIC GENETIC ALGORITHM FOR DIOPHANTINE EQUATIONS SOLVING

  Е.Е. Polupanova, P.E. Usov

  115-123

  2022-01-31

  Abstract ▼

  The problem of diophantine equations solving is considered in this article. This problem can
  be applied in cryptography and cryptanalysis. The description of the genetic algorithm solving
  diophantine equations is stated briefly in the article. The rule of calculation the value of fitness
  function of chromosome is determined, the coding system in the genetic algorithm is described.
  The genetic operators used in the algorithm are mentioned and the conditions for their execution
  are determined. The criterion for stopping the genetic algorithm is described. One of the shortcomings
  of the genetic algorithm is analyzed. The shortcoming of the algorithm lies in its attempts
  to solve any diophantine equation, including one that has no solutions. A method eliminating this
  shortcoming in some cases is proposed. This method is based on number theory. An explanation is
  given in which cases this method will be used. The definition of residue and nonresidue of fixed
  power for fixed modulus is given before describing this method. After describing this method the
  implementation of the algorithm for solving diophantine equations and systems of them is described
  in detail. Then the results of experimental studies of the time and quality of the genetic
  algorithm are presented. Then the result of the algorithm is presented for an equation that has no
  solutions and for a system of equations that also has no solutions, but in which the total number of
  unknowns is too large for the proposed method to work. The algorithm running time is compared
  when solving an equation and when solving a system of equations. The conclusion is made about
  the usefulness of the proposed method in solving diophantine equations and systems of diophantine
  equations.

• METHOD FOR SEARCHING SEQUENTIAL PATTERNS OF USER'S BEHAVIOR ON THE INTERNET

  V.V. Kureychik, V. V. Bova, Y.A. Kravchenko

  2020-11-22

  Abstract ▼

  One of the important tasks of data mining is to isolate patterns and detect related events in
  sequential data based on the analysis of sequential patterns. The article examines the possibility of
  using sequential patterns to analyze the events of search and cognitive activity of users when interacting
  with Internet resources of an open information and educational environment. Searching
  for sequential patterns is a complex computational task whose goal is to retrieve all frequent sequences
  representing potential relationships within elements from a transactional database of
  sequences of search activity events with a given minimum support. To solve it, the article proposes
  a method for searching for patterns in sequences of events to detect hidden patterns that indicate
  possible levels of vulnerability when performing information search tasks in the Internet space.
  A mathematical model of user behavior in a search session based on the theory of sequential patterns
  is described. To improve the computational efficiency of the method, a modified algorithm
  for generating sequential patterns has been developed, at the first stage of which AprioriAll is
  performed, which forms frequent candidate sequences of all possible lengths, and at the second
  stage, a genetic algorithm for optimizing the input parameters of the feature space of the generated
  set to search for maximum patterns. A series of computational experiments were carried out on
  test data from the MSNBC corpus, the SPMF open source data mining library. The comparative
  analysis was carried out with the VMSP and GSP algorithms. The research results confirmed the
  efficiency of the search for maximum sequential patterns by the proposed algorithm in terms of the
  execution time and the number of extracted patterns. The results of the experimental studies of the
  method showed that to increase the stability and accuracy of the work, the sample size obtained as
  a result of the GA operation will reduce the required number of scans of the pattern database,
  providing acceptable computational costs comparable to the VMSP algorithm and the GSP algorithm
  that exceeds the search time for sequential patterns. an average of more than 150 %.

• A GENETIC ALGORITHM FOR PLANNING THE TRAJECTORY OF A GROUP OF MOBILE ROBOTS IN THE PRESENCE OF STATIONARY AND MOBILE OBSTACLES

  L. А. Rybak, D.I. Malyshev, D. А. Dyakonov, А. А. Mamchenkova

  2025-04-27

  Abstract ▼

  The article discusses a trajectory planning method for a group of mobile robots that ensures safe
  movement and eliminates the possibility of collisions both between the robots themselves and with external
  obstacles, including moving objects. The developed mathematical model considers three main collision
  scenarios: intersection of robot trajectories within the group, interaction with stationary obstacles, and the probability of collision with moving objects. Each of these scenarios is analyzed in detail to ensure
  maximum safety during movement, and their consideration allows for efficient adaptation of robot routes
  to changing environmental conditions. The trajectory of each robot is represented as a piecewise linear
  path with intermediate points, which are optimized to ensure safe movement. Special attention is paid to
  speed adaptation on different segments of the trajectory: a robot can adjust its speed based on current
  conditions to minimize the risk of collisions. To evaluate distances between objects, the Euclidean norm is
  used, allowing for the calculation of minimum distances between the centers of spherical representations
  of robots and obstacles. The problem is solved in two stages. In the first stage, a trajectory is constructed
  for the first robot, taking into account initial conditions and obstacle placement. In the second stage, trajectories
  are formed for the remaining robots, considering the already planned routes. For optimizing the
  coordinates of intermediate points and speeds, a genetic algorithm is applied, which minimizes travel time
  while ensuring safe movement. The genetic algorithm uses crossover and mutation operators to generate
  diverse solutions and performs checks to ensure compliance with safety conditions. Numerical simulations
  were conducted using Python, with the Matplotlib library used for visualization of results. During the
  experiments, 50 tests were performed with varying numbers of obstacles (from 5 to 10). Analysis of the
  results showed that as the number of obstacles increased, both the computation time and the quality of the
  generated trajectories improved. This confirms the effectiveness of the proposed method for controlling
  groups of mobile robots in dynamically changing environments

• ESTIMATING THE EFFECTIVENESS OF THE METHOD FOR SEARCHING THE ASSOCIATIVE RULES FOR THE TASKS OF PROCESSING BIG DATA

  V. V. Bova, E.V. Kuliev, S.N. Scheglov

  2020-07-20

  Abstract ▼

  The modern databases have significant volume and consist of large masses of information.
  One of the popular methods of knowledge identification in terms of tasks of analysis and processing
  of large data volumes is composed of the algorithms for searching the associative rules.
  The paper solves the problem of building the bases of associative rules for the analysis of the unstructured
  large data volumes on the basis of searching different regularities considering the importance
  of their characteristics. The authors propose the method for synthesizing the bases and
  building the transaction database to calculate the threshold values of support and application of
  criteria of estimating implicit associations. This allows us to extract repeated and implicit associative
  rules. To improve the computational effectiveness of extracting the associative rules, the paper
  applies the genetic algorithm for optimization of input parameters of the characteristic searching
  space. The developed method shortens the time of rules extraction, reduces the number of generated
  common rules, and avoid the resource-consuming procedure of pre-processing the synthesized
  rule base. The authors developed the program and algorithmic module to carry out the experimental
  research of the proposed method for synthesizing the associative rules on the basis of filtering
  the input parameters of the search model for solving the tasks of processing the unstructured
  data. The experiments conducted on the test transaction bases allow us to clarify the theoretical
  estimations of time complexity of the proposed method that used the genetic algorithm to calculate
  the weighed support of the set of rules considering the assessment of a priori informative content
  of the characteristics included in the dataset. The time complexity of the developed method is estimated
  as  О(I2). The comparative analysis is performed using the test data of the Retail Data
  with the algorithms Apriori and Frequent Pattern-Growth. The results have proven the effectiveness
  of the search method on big sets of transactions. The method allows us to reduce the cardinal
  of an irredundant set of extracted associative rules in more than 40% in comparison with the popular
  algorithms. The experiments have shown that the method can be effective for the tasks of
  knowledge discovery in terms of processing large volumes of data.

• MODIFIED GENETIC PROJECT PLANNING ALGORITHM IMPLEMENTED WITH THE USE OF CLOUD COMPUTING

  А. А. Mogilev, V.M. Kureichik

  2020-07-20

  Abstract ▼

  The paper proposes a structure of a modified genetic algorithm for solving resource constrained
  project scheduling problem implemented with the use of cloud computing, a computational
  experiment was conducted, during which the results of the proposed algorithm were compared
  with the best known, at the moment, results. Based on the results of the experiment, it was concluded
  that the proposed algorithm can be used to plan the work of real projects, since it is possible
  to draw up schedules for projects with the number of works n = 90 for an acceptable period of
  time. When planning projects with the number of jobs n = 30, n = 60, n = 90, 120, the execution
  time of the proposed algorithm was less than the execution time of the standard genetic algorithm
  by 2.8, 4, 5.5 and 6.8 times, respectively. Due to the fact that the task of constructing a project
  schedule taking into account limited resources is NP-difficult, the problem of creating new and
  modifying existing methods for solving it remains relevant. For planning projects with a large
  number of works, it is advisable to use cloud computing, since planning such projects can require
  a lot of time and computing resources. In this regard, the algorithm proposed in this paper differs
  from the existing ones by using cloud computing to distribute the load between workstations on
  which this algorithm is simultaneously running. The use of modified operators in the genetic algorithm,
  as well as the use of cloud infrastructure as a service for implementing a distributed genetic
  algorithm, determines the scientific novelty of the study.

• CLASSIFICATION AND ANALYSIS OF EVOLUTIONARY METHODS OF EVA BLOCK LAYOUT

  Y.V. Danilchenko, V.I. Danilchenko, V. M. Kureichik

  2020-07-20

  Abstract ▼

  Currently, there is a large increase in the need for the design and development of radioelectronic
  devices. This is due to increasing requirements for radio-electronic systems, as well as
  the emergence of new generations of semiconductor devices. In this regard, there is a need to develop
  new tools for automated layout of EVA blocks. There are a number of problems that complicate
  the actual representation of knowledge in CAD and are probably solvable at the current level
  of cognitive science development. The problem of stereotyping and the problem of coarsening are
  interrelated and need to create hybrid models of representation. The paper deals with the problem
  of solving the problem of EVA block layout in the design of radio-electronic equipment. The purpose
  of this work is to find ways to optimize the planning of EVA block layout using a genetic
  algorithm. The relevance of the work is that the genetic algorithm can improve the quality of layout
  planning. These algorithms allow you to improve the quality and speed of layout planning. The
  scientific novelty lies in the search and analysis of effective methods for composing EVA blocks
  using genetic algorithms. The main difference from the known comparisons is in the analysis of
  new promising algorithms for composing EVA blocks. Result of work. The paper shows the disadvantages
  of traditional algorithms for searching for a suboptimal EVA plan. Descriptions of modern
  models of evolutionary and other calculations are given. Genetic algorithms have a number of
  important advantages – adaptability to a changing environment, the evolutionary approach makes
  it possible to analyze, Supplement and change the knowledge base depending on changing conditions,
  as well as quickly create optimal solutions. If you apply genetic algorithms and preprocessing
  heuristics to provide optimal initial solutions, you can achieve more productive use of
  algorithms. Known genetic algorithms converge quickly, but they lose population diversity, which
  affects the quality of the solution. To balance data, the solution is corrected using efficient operators
  or stable mutation.

• ROBOT PATH PLANNING FOR MULTI-TARGETS BASED ON A HYBRID OF PRM AND AGA ALGORITHM

  Alzubairi Shaymaa М. Jawad Kadhim , А.А. Petunin , S.S. Ukolov

  6-18

  2025-11-10

  Abstract ▼

  Optimal path planning problems for mobile robots have been particularly actively studied in the last decade. The goal is to find an optimal or near-optimal path from a starting terminal to one or more terminals in an environment with various obstacles, in terms of minimizing robot travel time, distance traveled, energy costs, or other optimization criteria. In this paper, we propose a hybrid algorithm combining a probabilistic roadmap algorithm (PRM) and an adapted genetic algorithm (AGA) to solve a path planning problem with one or more independent objectives. The robot's path length is used as an optimization criterion. Compared with existing approaches used in genetic algorithms (GAs), the proposed approach has two main differences. The first is the environment representation, which relies on image processing and morphological operations, which has proven to be a more efficient method than methods based on cellular representation. In particular, the proposed method eliminates the need to find a trade-off between accuracy and speed of processing geometric information. The second is a new tactic for creating an initial population of the genetic algorithm to accelerate convergence in the presence of multiple objectives. By leveraging the capabilities of a probabilistic roadmap algorithm. Another key feature of the algorithm's implementation is the appropriate (for the domain under study) selection of numerical parameters that determine the characteristics of all stages of the evolutionary strategy, including the time required to complete each stage. This applies in particular to the parameters of the mutation operator and the elite strategy. The proposed algorithm was tested on two real-world maps with varying levels of complexity. Its effectiveness was confirmed by comparison with path planning results for test maps obtained using a standard genetic algorithm and an ant colony optimization algorithm. Experimental results demonstrate that the hybrid algorithm expands the capabilities of a conventional genetic algorithm and finds rational path variants with the best objective function value for single and multiple objectives in significantly less time than other traditional GA implementations.

• INTELLIGENT METHODS OF PARAMETRIC FORECASTING AND OPTIMIZATION OF UAV TRAJECTORIES

  V.I. Danilchenko , V.V. Bova

  263-276

  2025-12-30

  Abstract ▼

  This paper examines the problem of intelligent parametric forecasting and trajectory optimization for unmanned aircraft systems (UAS) using evolutionary algorithms and machine learning methods. The relevance of the study stems from the multi-criteria and high complexity of UAS trajectory generation processes, as well as the need for accurate and timely assessment of its flight parameters. This is particularly important for ensuring the reliability, safety, and efficient performance of flight missions in UAS operating conditions, including scenarios related to the operation of critical infrastructure facilities. The objective of the study is to improve the accuracy of trajectory parameter diagnostics and the reliability of parametric forecasting of UAS trajectories under conditions of uncertainty and the multi-criteria nature of the problem. The paper proposes a hybrid approach incorporating a genetic algorithm (GA), a particle swarm algorithm (PSO), and an XGBoost machine learning model that provides adaptive assessment of the quality of the generated solutions. A computational software package has been implemented, including selection, recombination, mutation, and elite inheritance mechanisms, as well as a machine learning module for validating route trajectories and associated parameters. A computational experiment was conducted, which compared the effectiveness of GA and PSO under various operating scenarios. Testing was performed on industry-specific datasets with varying numbers of iterations. The computational experiment revealed the advantage of the genetic algorithm, namely, a 14–17% improvement in the quality of design solutions. The results of the study demonstrate high adaptability and practical applicability in modeling, parametric forecasting, and routing tasks, and also indicate the potential for integration with intelligent UAS navigation and monitoring systems. The article's materials are of practical interest to specialists in the field of UAS development and operation, as well as to researchers working on multi-criteria route planning, parametric forecasting, and improving the reliability of UAS operations.

• ANALYSIS OF THE CAUSES OF ERRORS IN THE AMPLITUDE-PHASE DISTRIBUTION OF LINEAR PHASED ANTENNA ARRAYS AND METHODS FOR THEIR REDUCTION

  S.S. Bybin , N.P. Dunaev , S.V. Kuzmin , А.N. Morozov

  2026-02-27

  Abstract ▼

  To use a phased array antenna in the beamforming mode, it is necessary to establish a certain amplitude-phase distribution at the inputs of the emitting elements. Amplitude and phase errors distort the radiation pattern. The paper analyzes the sources of errors in the amplitude-phase distribution of phased antenna arrays, including parasitic phase shifts, nonlinear amplification paths, temperature instability and mutual electromagnetic coupling between the elements. Three methods of calibration and adjustment of phased antenna arrays are described, based on direct measurements of the transmission coefficients in the near zone and the subsequent calculation of the impact vector using inverse and pseudo-inverse matrices of mutual connections, which provides a systematic approach to error elimination. To obtain the initial values, each channel was pre-calibrated along a closed path using a vector network analyzer. Technique 1 implements correction for a set of points in space and one set of states of each channel. To increase the stability of the solution, method 2 uses the regularization of the elements of the matrix of interconnections based on an additional set of measured states of each channel. Method 3 makes it possible to construct a mathematical model of a specific implementation of a phased array antenna based on measurements with a fixed channel state, which ensures the formation of an arbitrary amplitude-phase state without repeated measurements. An experimental setup of an eight-element equidistant linear phased array antenna was carried out. The lattice attenuator/phase shifter modules are based on the PE44820 phase shifter and PE4302 attenuator debugging boards and are controlled by a microcontroller to automatically change phases and amplitudes. The measurements were carried out automatically on a near-field stand in an anechoic shielded chamber using a vector network analyzer. Calibration results are presented, matrices of mutual relationships are constructed and radiation patterns are formed, confirming the operability of the proposed approaches. Since the experimental array is low-element, the results of applying the considered techniques are compared with the results of tuning in the far zone performed using an evolutionary algorithm.

• PREDICTING BOND PRICE MOVEMENTS USING A HYBRID METHOD BASED ON XGBOOST AND A GENETIC ALGORITHM

  L. E. Khairullina , Z.N. Khakimov , D.I. Galiev , А.N. Khairullina

  208-219

  2026-07-07

  Abstract ▼

  The article presents a hybrid method for predicting the direction of bond price movements, combining the XGBoost machine learning method with hyperparameter optimization using a genetic algorithm. The research is aimed at solving the problem of binary classification of the direction of the price of the Russian Railways bond on the next trading day. The research methodology includes the formation of an expanded feature space of 18 technical indicators calculated on the basis of daily OHLCV data. To configure XGBoost hyperparameters, a genetic algorithm is implemented using the DEAP library. The study was conducted on three time horizons: 01.01.21-31.10.25, 01.01.23-31.10.25, from 01.01.24-31.10.25.
  As a result, a significant dependence of the effectiveness of the model on the time horizon of the training data is shown. The best quality was demonstrated by a model trained on data from 2024-2025, with an accuracy of 64.4% in the test sample, balanced precision and recall metrics, as well as high F1-score scores for both classes. Models trained over longer periods (2021-2025 and 2023-2025) showed a decrease in generalizing ability, which indicates that the relevance of the data prevails over its volume in the context of changes in Russia's monetary policy in 2021-2025. To maintain the predictive power of the model in changing market conditions, it is recommended to use a sliding learning window of 1.5–2 years. The comparison with the "Buy & Hold" strategy confirmed the effectiveness of the proposed hybrid approach