For authors
Search
Search Results
##search.searchResults.foundPlural##
-
A MODEL OF RESOURCES ALLOCATION INFORMATION PROCESS IN DYNAMIC DISTRIBUTED COMPUTING ENVIRONMENTS
А.B. Klimenko110-1202025-10-01Abstract ▼The article considers the issue of modeling the information process of distributing computing resources in geo-distributed heterogeneous dynamic computing environments. The relevance of the work is due to the fact that by now "cloud" data processing systems are becoming insufficient due to the need to process large volumes of data in real time regime. In this regard, the "fog" and "edge" computing are in use. This implies localization of data processing in order to reduce the time required for this, on the one hand, and on the other hand, limitations on the computing power of devices leads to the need for a distributed solution of computing problems in a heterogeneous, dynamic and geographically distributed environment. This entails the need to develop new methods and algorithms for computing resources allocation, since previously developed methods did not take into account the properties of geographic distribution and dynamics of computing environments. The model of the information process of computing resources allocation proposed in this work includes the parameters of the resource cost of data transfers over the network individually for the nodes participating in the data transfer route, as well as the process of distribution of computing resources, which is what distinguishes it from analogs. The conducted experimental studies confirm the feasibility of the proposed model usage for the computing resources allocation in geo-distributed heterogeneous dynamic computing environments. The practical significance lies in reducing the resource intensity of the process of distribution of computing resources and the process of solving a computing problem
-
IMPLEMENTATION OF CONVENTIONAL NEURAL NETWORKS ON EMBEDDED DEVICES WITH A LIMITED COMPUTING RESOURCE
V.V. Kovalev, N.E. Sergeev2022-01-31Abstract ▼Large amounts of video data captured by sensor sensors in various spectral ranges, the significant
size of convolutional neural network architectures create problems with the implementation of
neural network algorithms on peripheral devices due to significant limitations of computing resources
on embedded computing devices. The article discusses the use of algorithms for automatic search and
pattern recognition based on machine learning methods, implemented on embedded devices with a
computing resource Graphics Processing Unit. Detection convolutional neural networks «You Only
Look Once V3» and «You Only Look Once V3-Tiny» are used as a search and pattern recognition algorithm,
which are implemented on embedded computing devices of the NVIDIA Jetson line, located in
different price ranges and with different computing resources ... Also, in the work, the estimates ofalgorithms on embedded devices are experimentally calculated for such indicators as power consumption,
forward passage time of a convolutional neural network, and detection accuracy.
On the basis of solutions implemented, both at the hardware level and in software, presented by
NVIDIA, it becomes possible to use deep neural network algorithms based on the convolution
operation in real time. Computational optimization methods offered by NVIDIA are considered.
Experimental studies of the influence of computations with reduced accuracy on the speed and
accuracy of object detection in images of the investigated architectures of convolutional neural
networks, which were previously trained on a sample of images consisting of the PASCAL VOC
2007 and PASCAL VOC 2012 datasets, have been carried out. -
THE USE OF HETEROGENEOUS COMPUTING NODES IN GRID SYSTEMS IN SOLVING COMBINATORIAL PROBLEMS
А.М. Albertian, I. I. Kurochkin, E.I. Vatutin142-1532021-10-05Abstract ▼The main goal of this work is to create a parallel application that performs computations using a multithreaded execution model, optimized to make the best utilization of all available hardware resources. One of the main implementation requirements is to optimize application per-formance on different computer architectures, and to enable parallel execution of the application on various computing devices that are part of a heterogeneous computing system. The possibility of applying various methods of software and algorithmic optimization on multiprocessor architec-tures of different generations was investigated as well as the effectiveness of their use for highly loaded multithreaded applications was estimated. The problem of quasi-optimal dynamic distribu-tion of computational tasks among all currently available computing devices of a heterogeneous computing system was also solved. Currently, not only multiprocessor computing systems are used to solve large computational problems, but also various types of distributed systems. Distributed computing systems have a number of features: possible failures of nodes and communication channels, unstable operating time of nodes, possible errors in calculations, heterogeneity of com-puting nodes. By heterogeneity of computing nodes, we will understand not only the different com-puting capacity and different architectures of central processors, but also the presence of other devices on the node capable of performing calculations. Such devices include video cards and mathematical coprocessors. A node of a distributed computing system will be called heterogene-ous if, in addition to one or more central processing units, it contains additional computing devic-es. When solving a computational problem on a distributed system, it is necessary to maximize the utilization of all available computing resources. To do this, it is necessary not only to distribute computing subtasks to nodes in accordance with their computing capacity, but also to take into account the features of additional computing devices. This work is devoted to the study of methods for maximizing the resources utilization of heterogeneous nodes.
-
METAHEURISTICS BASED ON THE BEHAVIOR OF A COLONY OF WHITE MOLES
Y.V. Danilchenko, V. I. Danilchenko, V. М. Kureichik132-1402021-08-12Abstract ▼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
-
A NEW ALGORITHM FOR CONSTRUCTING THE SHORTEST TOUR OF A FINITE SET OF DISJOINT CONTOURS ON A PLANE
А. А. Petunin, E.G. Polishchuk, S.S. Ukolov2021-04-04Abstract ▼The problem of tool path routing for the CNC thermal cutting machines is considered.
Pierce points are located at the parts bounding contours, consisting of straight-line segments and
circular arcs. Continuous cutting technique is used, each contour is cut out entirely, and no presampling
occurs, so cutting can start from any point on the contour. General problem of minimizing
the route length is reduced to minimizing the air move length. It is shown to be equivalent to
finding the shortest polyline with vertices on the contours. New algorithm for constructing such a
broken line for fixed order of contour traversing is proposed. The resulting solution is shown to be
a local minimum. Some sufficient conditions are described for the it to be also a global minimum,
which can be easily verified numerically, and some even visually. A technique is described for
automatically taking into account precedence constraints for the practically important case of
nested contours. This also decreases the size of the problem, which has a positive effect on the
optimization time. A heuristic routing algorithm based on the variable neighborhood search (VNS)
is proposed. Alternative approaches to the use of other discrete optimization methods along with
the proposed algorithm for constructing the shortest polyline for solving the complete problem of
continuous cutting, and the resulting difficulties of both theoretical and practical nature are described.
The generalization of the problem of continuous cutting to a wider class of problems of
(generalized) segment cutting is described, which makes it possible to advance in solving the problem
of intermittent cutting. The scheme of application of the proposed algorithm for solving problems
of generalized segment cutting is described. The results of numerical experiments are considered
in comparison with the exact solution of the GTSP problem. -
SOLUTION OF THE PROBLEM OF INTELLECTUAL DATA ANALYSIS BASED ON BIOINSPIRED ALGORITHM
E.V. Kuliev, D.Y. Zaporozhets, Y.A. Kravchenko, М.М. Semenova2022-01-31Abstract ▼The article discusses a bioinspired algorithm for solving the problems of intellectual analysis.
The integration of bioinspired algorithms for solving data mining problems is a promising
area of research. As a bioinspired algorithm, an algorithm based on the adaptive behavior of an
ant colony is considered. The ant colony algorithm allows for a high-quality search for promising
solutions to obtain optimal and quasi-optimal solutions. The algorithm has the ability to search for
suitable logical conditions. The ant colony algorithm is based on the example of the behavior of
living ants in nature. Ants are able to find the shortest solution by adapting to changes in the environment.
The authors proposed a modified ant colony algorithm for solving the problem of data
mining. The clustering problem was chosen as the task of data mining. Clustering is a combining
of similar objects into groups, is one of the fundamental tasks in the field of data analysis and
Data Mining. The list of application areas where it is applied is wide: image segmentation, marketing,
anti-fraud, forecasting, text analysis and many others. The solution to this problem is of particular relevance in the context of the constantly growing volume of generated, transmitted and
processed data. Classical clustering methods are optimized by combining with the proposed
bioinspired optimization algorithm - the ant algorithm. The proposed method is a model in which
ants are represented as agents that randomly move in the solution space with some restrictions
(for example, obstacles in their path). To determine the effectiveness of the developed modified ant
algorithm (ALA) with the clustering algorithm, the authors carried out a series of computational
experiments. For comparison, we took the genetic algorithm, the monkey algorithm and the wolf
algorithm. The simulation results prove that the clustering-based ant algorithm gives better results
than other proposed algorithms. -
HYBRID EXECUTION OF QUERIES TO ANALYTICAL DATABASES
P. A. Kurapov2021-07-18Abstract ▼Analytical database engines should benefit from evolving heterogeneous distributed architectures
and utilize their resources efficiently: various accelerators, complex memory hierarchy,
and distributed nature of systems bring performance improvement opportunities. The article reviews
existing approaches for in-memory DBMS query executor implementation using hardware
accelerators, especially GPUs. Massive parallelism and high on-device memory bandwidth make
graphics processors a promising alternative as a core query evaluation device. Existing methods
do not utilize all modern hardware capabilities and usually are bound, performance-wise, by relatively
slow PCIe data transfer in a GPU-as-a-co-processor model for each kernel representing a
single relational algebra operator. Another existing approaches’ issue is explicit code base separation
for relational algebra operators code generation (for CPU and GPU) that significantly
limits possibilities of joint device usage for performance increase and make it less feasible. The
article presents an efficient query execution method using an example of two device classes (CPU
and GPU) by compiling queries into a single, device agnostic, intermediate representation (SPIRV)
and an approach for corresponding hybrid physical query plan optimization based on extended
classical “Exchange” operator with explicit control over heterogeneous resources and parallelism
level available. A cost model composition process using basic compute DBMS patterns benchmarking
and buses bandwidth data for both relational and auxiliary operators is proposed. Potential
processing speedup from holistic query optimization is estimated empirically with a commercial
open source DBMS OmniSci DB. Preliminary results show significant (3-8x, depending on
device choice) possible speedup even with just using the right device for the job. -
MODELING OF THE VACUUM INFUSION PROCESSES IN THE MANUFACTURING OF THE LARGE POLYMERIC COMPOSITE STRUCTURES
Huang Jyun-Ping2021-08-11Abstract ▼The article presents the technology of computer simulation of the vacuum infusion process
in the production of large-sized polymeric composite structures, which is attracting more and
more attention in the aircraft industry, due to the ease of implementation and the relatively low
cost of production preparation. The difficulty of industrial implementation of the process and ensuring
the required quality is due to its high sensitivity to modes - temperature, vacuum pressure
and the layout of the vacuum ports and resin injection. The purpose of the developed methodology
for computer modeling of the process with the possibility of its subsequent optimization is to exclude
the currently used lengthy and very expensive trial and error method when working out the
technology. The proposed mathematical model of the process linking the equation of the phase
field, which reconstructs the interface between the resin and the void region of the preform, the
Richards equation for the propagating viscous fluid in an unsaturated porous medium, the thermal
kinetics of the resin and thermal conductivity, is implemented in the environment of a finite element
package. Computer implementation of the model provides an accurate reconstruction of the
dynamics of the front of the propagating resin in a porous preform, the possibility of the emergence
and localization of non-impregnated zones of the molded structure, thereby eliminating the
formation of irreparable defects. The results obtained demonstrate the ability of the developed
technique to ensure the stability of the quality of the produced composite structures with increased
requirements for the continuity of its microstructure and its structural strength. -
OPTIMIZATION-BASED CALIBRATION OF MEMS NAVIGATION SYSTEM
D.E. Chickrin , S.V. Golousov2021-08-11Abstract ▼Technologies of autonomous wheeled robotic systems are becoming more and more in demand
lately. A separate type of application of such technology is an autonomous unmanned
ground vehicle. Unlike other types of transport (air, water), ground vehicles need to periodically
operate in full autonomy - when external communication with the infrastructure and other agents
of the transport network is inaccessible. In such circumstances, the issue of autonomous navigation
comes out on top, and increased requirements are imposed on positioning accuracy, especially
in an anthropogenic environment, for example, when driving in an urban environment, along
narrow mountain roads, and tunnels. One of the components of autonomous navigation is often an
inertial assembly consisting of several accelerometers, gyroscopes, and magnetometers. To obtain
a high-precision navigation solution based on an inertial assembly, it is required to properly calibrate
it. A separate issue is automation and its cost for further scaling necessary for mass production.
The article presents the theory and methodology for automated calibration of an inertial
navigation system based on MEMS sensors by solving an optimization problem. The proposed
technique does not require high-precision calibration equipment. The aim of the presented work is
to develop methods and theory for the calibration of inertial navigation units. The article formulates
general measurement models of sensors included in the inertial assembly, and proposes
methods for calibrating the parameters of accelerometers and gyroscopes fixed relative to each
other. The method of automation of the calibration process is presented, which does not require
high-precision equipment. The results of the application of the developed methods for the calibration
of a real inertial assembly are presented. A stand for automated calibration is presented. -
HYBRID METHOD FOR SOLVING THE PROBLEM OF PLACEMENT OF DIGITAL COMPUTER DEVICES
L. A. Gladkov , N. V. Gladkova , M.J. Yasir2021-11-14Abstract ▼The problem of placing elements of digital computing technology is considered in the article.
The analysis of the current state of research on this topic is carried out, the relevance of the
problem under consideration is noted. The importance of developing new effective methods for
solving such problems are highlighted. The place of the placement problem in the general cycle ofthe design stage is shown. The importance of a high-quality solution to the placement problem
from the point of view of the successful implementation of subsequent design stages is noted. The
importance of minimizing connection delays in the design process of large-scale devices is noted.
A review and analysis of various models and criteria for evaluating the solution to the placement
problem is carried out. It was emphasized that the most important criterion is the length of the
joints, it has a significant impact on the technologies used in the design. A complex mathematical
formulation of the problem of placing elements of digital computing equipment has been completed.
Perspective approaches to solving design problems are analyzed, hybrid methods and models
for solving complex multicriteria optimization and design problems are described. The principles
of operation and the model of a fuzzy logic controller are described. The description of the used
fuzzy control scheme is given. The functions of various blocks of a fuzzy logic controller are determined.
The structure of a multilayer neural network that implements the Gaussian function is
proposed. The interaction of blocks of a fuzzy genetic algorithm is described. A model of a hybrid
algorithm for solving the placement problem is proposed. The control parameters of the fuzzy
logic controller are determined. The proposed hybrid algorithm is implemented as an application
program. A series of computational experiments to determine the effectiveness of the developed
algorithm and select the optimal values of the control parameters were carried out. -
METHOD OF AUTOMATIC OPTIMIZATION OF THE FUZZY RULE BASE OF AN INTELLIGENT CONTROLLER BASED ON SUBTRACTIVE CLUSTERING
А.S. Ignatyeva , V.V. Shadrina , D.S. Ignatyev , А.V. Maksimov181-1972025-07-24Abstract ▼The aim of the work is to develop a method for optimizing the fuzzy rule base of an intelligent controller for controlling a technical object using subtractive clustering. The article provides an overview and a brief analysis of the state of affairs in the field of optimizing the operation of intelligent control systems. To achieve the goal of the study, a hybrid model has been developed in which the technical object is controlled using a classical PI controller and a fuzzy PI controller with a generated structure of a Cygeno-type fuzzy inference system and a developed model of an adaptive neuro-fuzzy inference system. This configuration of the model allows you to form a fuzzy rule base that does not depend on the expert's knowledge in the subject area. The article proposes a new method for optimizing the fuzzy controller rule base based on clustering methods, in particular subtractive clustering, which allows you to reduce the number of fuzzy logical inference rules and increase the performance of the technical object control system. First, a hybrid model synthesized on the basis of the values of the fuzzy and classical controllers before applying subtractive clustering was simulated. The application of subtractive clustering according to the method developed in the study for the values of the classical and fuzzy controllers allowed us to achieve their quantitative reduction by 1.7 and 5.25 times, respectively. Then, the hybrid model synthesized on the basis of the values of the fuzzy and classical controllers after applying subtractive clustering was simulated. The results obtained in the process of simulation showed high efficiency of the proposed method for optimizing the fuzzy controller rule base. Due to the application of subtractive clustering in the hybrid model for the intelligent controller, it was possible to significantly reduce the number of membership functions required to describe the input linguistic variables (from five to four) and reduce the number of fuzzy logical inference rules (from twenty-five to sixteen). The analysis of the resulting graphs of transient processes obtained for the hybrid models before and after applying subtractive clustering showed that the main indicators of the quality of the control process remain unchanged with a significant reduction in the calculations performed.
-
ALGORITHM OF EFFECTIVE CONTROLS FOR NONSTOCHASTIC CAUSAL MODELS IN THE ABSENCE OF OBSERVABLE VARIABLES FOR SYSTEMS OF DECISION MAKING CONTROL
A.N. Tselykh, V.S. Vasilev , L.A. Tselykh2021-11-14Abstract ▼The paper deals with the problem of reproducing the decision-making process by a person under
conditions of uncertainty and incompleteness of the initial data. The decision-maker relies on his
belief system, which includes a shared vision of the system in relation to which the decision is being
made. The system is presented in the form of a causal model created on the basis of human mental
representations. These models are directed graphs, on the arcs of which the causal relationship is
expressed in the form of labels with a sign that determines the direction of change in the state of the
system. The vertices of this directed graph are high-level abstraction concepts. This graph simulates
the functioning of a real system. Thus, we investigate the problem of predicting and controlling human
actions based on non-stochastic causal models in the absence of observable variables for use in
decision support systems and expert systems. Decision-making is considered from the point of view of
the choice of objects of application of managerial influences - the factors of the model. In this study,
we show that the application of the proposed algorithm can facilitate decision-making regarding the
choice of control actions that support the achievement of the tactical and strategic goals of the decision
maker. It should be noted that the algorithm implements an automatic selection of the regularization
parameter, which makes the development and application of the proposed algorithm available
to users who do not have sufficient mathematical training. The convergence of the sequence of Lagrange
multipliers of an effective control algorithm is proved. The theorem on resonance in a nonstochastic
causal mod-el, represented by a directed graph, which is determined by the range of admissible
values of the damping coefficient in the control model, is proved. It is expected that the introduction
of this tool into decision support systems will in-crease the reliability of decisions regarding
the operation of the system as a whole. The choice of control actions using the proposed algorithm
has high efficiency and productivity. Thus, the results presented in the study can be useful for
developing applications in intelligent systems. -
DEVELOPMENT OF MODIFIED METHODS AND MODELS OF SEARCH ADAPTATION FOR SOLVING THE PROBLEM OF PLANNING VLSI
O.B. Lebedev, А.А. Zhiglatiy, Е.О. Lebedevа2021-12-24Abstract ▼In this work, to solve the VLSI planning problem, a search algorithm has been developed
based on a modified ant colony method. The task of forming a VLSI plan is reduced to the task of
forming the corresponding Polish expression. The developed method for the synthesis of the Polish
expression includes the construction of a tree of cuts, the choice of the types of cuts (H or V), identification
and orientation of modules. The evolving population is split into pairs of agents. Each
member of the population is a pair of agents working together. In this case, the constructive algorithms
A1 and A2 used by the agents of the pair are different. The problem solved by Algorithm A1
is formulated as the problem of finding a one-to-one mapping Fk=M*→P of the set of modules M
with selected orientations, |M*|=|M| to the set P of positions of the template Sh. In fact, the solution
consists in choosing on the graph G1 a subset of edges E*1E1 included in the corresponding
mapping Fk. In Algorithm A2, the graph G2=(X, E2) is developed as a model of the search space
for solutions for choosing the type, sequence and location of cuts in the pattern Sh.
X={(x1i,x2i)|i=1,2,…,n} the set of vertices of the graph G2, corresponds to the set P of potential
positions of the template Sh for the possible placement of the names of the cut symbols in them.
Each potential position piP of the template Sh is modeled by two alternative vertices (x1i,x2i).
The choice of the vertex x1i when placing the cuts indicates that a cut of type V is placed in position
pi, the choice of vertex x2i indicates that a cut of type H is placed in position pi. Each iteration
l of the general algorithm includes an initial and three main stages. The initial stage is as follows.
Co-evolutionary memory matrices are nullified CEM*1 and CEM*2 are reset to zero. At the first
stage, each pair of agents dk=(a1k,a2k): – with constructive algorithms A1 and A2 he synthesizes
his solution Wk=(E1k
*,Sk); – the Polish expression Shk is formed, corresponding to the solution Wk;
– on the basis of Shk, a tree of sections Tk is formed; – on the basis of Tk, the plan Rk is formed and
the estimate of the solution Fk is calculated; – agents deposit (add) the pheromone to the cells of
the collective evolutionary memory (CEM) matrices CEM*1 and CEM*2 corresponding to the
solution edges Wk=(E1k
*,Sk) in the solution search graphs G1 and G2 in an amount proportional
to the solution estimate Fk. At the second stage, the pheromone accumulated in CEM*1 and
CEM*2 by agents of the population at iteration l is added to CEM 1 and CEM2. At the third stage,the pheromone is evaporated on the edges of the graphs G1 and G2. Tests have confirmed the
effectiveness of the proposed method. The time complexity of the algorithm, obtained experimentally,
coincides with theoretical studies and it is O(n2) for the considered test problems. -
ACCELERATION OF THE DIRECT PASSAGE IN THE IMPLEMENTATION OF CNN ON A LIMITED COMPUTING RESOURCE
А.Е. Shchelkunov, V.V. Kovalev, I. V. Sidko, N. Е. Sergeev2022-04-21Abstract ▼The work is devoted to the optimization of the neural network architecture for its launch on
a limited computing resource. Several optimization approaches are considered, estimates of the
complexity and execution time of the forward pass of the neural network are given. Comparative
estimates of the complexity of the network using different optimization approaches are given.
The paper presents an analysis of the selected network architecture, and estimates of the computational
complexity of individual components (modules) of the architecture are obtained. An analysis
of possible optimization methods for each module was made. The parameters of the considered
modules, the sizes of the input and output tensors are described. Several architectures were tested
to optimize the feature extraction module, ResNet 50, ResNet 18, MobileNet v3 small, MobileNet
v3 large. A comparative analysis of the computational complexity and execution time of the forward
pass for each architecture is presented. Forward pass times were measured on Nvidia's
Jetson AGX Xaver embedded computing device. Estimates of the execution time of the direct pass
for each module of the considered neural networks are presented. The paper presents the results of
comparing neural network accuracy estimates before and after architecture optimization. The test
data set consists of 100 video recordings. 5 different typical objects are involved in test videos,
10 different scenarios are recorded for each object class. For each of the developed architectures,
accuracy estimates were obtained, and a comparative analysis was made. -
METAHEURISTIC OPTIMIZATION METHOD BASED ON THE STEM CELL BEHAVIOR MODEL
Y. V. Danilchenko , V.I. Danilchenko, V.M. Kureichik2022-05-26Abstract ▼The paper discusses optimization methods that are based on processes occurring in nature. Such
methods have become increasingly used to solve complex problems. However, such methods have some
drawbacks, which stimulates the development of new and more advanced optimization methods. Solving
NP complete problems requires optimal methods that will meet all design requirements, so there is a
need to develop new and more advanced methods for solving this class of problems. As such a method,
the authors propose an optimization method based on a model of the behavior of stem cells in the natural
environment. The conducted studies of the proposed method provide solutions that can overcome
many of the shortcomings of standard optimization approaches, such as getting into the local optimum
or low convergence rate of the algorithm based on the method under consideration. The purpose of this
work is to develop an optimization method and an algorithm based on it for solving a complex objective
function. The scientific novelty lies in the development of an optimization method based on the stem cell
behavior model for solving NP complete problems. The aim of the work is to create conditions for theoptimal search for a solution to complex functions by applying the search method and, based on it, an
algorithm for the behavior of stem cells. The practical value of the work lies in the development of a new
metaheuristic optimization method for the efficient solution of NP complete problems. Also in the work,
a comparative analysis with well-known competitors was carried out. The main difference of the proposed
method from other known methods is the use of a new approach of bioinspired search based on
the behavior of stem cells, which, as shown by practical comparison, has an advantage over known
analogues. The results of a practical comparison of methods and algorithms based on them showed the
advantages of the approach proposed in the work on known test functions. After analyzing the problem
of creating methods, algorithms and software for solving NP complete problems, we can conclude that
the development of such approaches is currently an urgent task. -
OPTIMIZATION OF PROJECT SCHEDULING UNDER UNCERTAIN PARAMETERS
А. V. Bozhenyuk, О. V. Kosenko, М.V. Knyazeva2022-05-26Abstract ▼This article considers the problem of operational planning of one-subject production.
The organization of machine-building production is a complex set of works to determine the interrelated
indicators that characterize the activities of the enterprise. Enterprises of this type have a
complex hierarchical structure. It is also necessary to take into account that when planning the
production process, the number of parameters is large and not all of them can be accurately determined,
which affects the efficiency of the enterprise. To solve the problem of effective planning,
the optimality criteria for serial one-subject production were analyzed. One-subject production
includes those where parts of the same name are processed, that is, a production line is formed.
Consequently, the task of optimizing production is to distribute the entire set of work between the
machines and operators servicing this machine in such a way that the planned task is completed
within a given time and the total cost of completing the task is minimal. The article considers the
problem of assignment under uncertainty, carried out experimental calculations and analyzed the
results obtained, which justifies the use of the proposed apparatus of fuzzy sets for solving the
problem of production planning. It is concluded that under conditions of uncertainty, when there is
no exact or statistical information, the apparatus of fuzzy sets makes it possible to analyze theeffectiveness of production activities when setting parameters that reflect the possible values of the
system. In such cases, the use of fuzzy logic mechanisms in the problems of making production
decisions will make it possible to determine optimal or close to optimal solutions. -
THE USE OF DISTRIBUTIVE SEMANTICS IN THE IDENTIFICATION OF SIGNIFICANT COMBINATIONS OF TITLES OF SEVERAL TEXT COLLECTIONS IN THE FORMALIZATION OF LINGUISTIC EXPERT INFORMATION
V.I. Danilchenko, V.M. Kureichik2022-08-09Abstract ▼The paper discusses methods of forming special models for the representation of various sets
of knowledge in various information systems. The work is devoted to the application of distributive
semantics in the identification of significant combinations in one subject area (PRO) within the
framework of the formalization of linguistic expert information (LEI). The paper applies an approach
to the formalization of LEI based on a set of analytical methods, where linear algebra is used as
models. This approach makes it possible to initialize the procedure for the automatic formation of
hierarchical architectures of LEI or dendrograms when identifying significant combinations of titles
of several collections of texts. The scientific novelty lies in the proposed analytical approach using
distributive semantics in identifying significant combinations of titles of several collections of texts,
which allows for the analysis and processing of linguistic expert information. A distinctive characteristic
of the proposed approach is the ability to formalize the ABM "Global Optimization Methods"
based on the synthesis of various already existing hierarchies of the ABM under consideration. The
paper aims to create conditions for the formalization of the LEI by applying distributive semantics
when identifying significant combinations of titles of several collections. The practical value of the
work lies in the development of a new approach to the formalization of LEI, taking into account distributive
semantics when identifying significant combinations of titles of several collections of texts.
The ontology in owl format "Methods of global optimization" in the program "Protege" is also built
in the work. The ontology is built on the basis of related data about. The ontology constructed in this
work complements the search structure within the framework of the considered PRO and can be
supplemented and developed in the future. -
STUDY OF PATH PLANNING METHODS IN TWO-DIMENSIONAL MAPPED ENVIRONMENTS
М. Y. Medvedev, V.K. Pshikhopov, D.О. Brosalin, B.V. Gurenko, М.А. Vasileva, Hamdan Nizar2022-08-09Abstract ▼The article studies the problem of motion planning in two-dimensional mapped environments.
The review and analysis of known planning algorithms based on Voronoi diagrams, probabilistic
road maps, rapidly growing random trees, Dijkstra algorithms, A*, D* and their modifications, artificial
potential fields and intelligent heuristics are carried out. Based on the analysis, it is concluded
that classical methods in dynamic environments require significant costs in terms of calculation time
and the amount of memory used. The conclusion is made about the relevance of the development of
algorithms that increase the efficiency of known planning methods. In this regard, this article is devoted
to the development of a modified algorithm of rapidly growing random trees and the study of its
effectiveness in comparison with known methods. The article presents a modified algorithm for rapidly
growing random trees, characterized in that when checking for a path to a new potential node of
the tree, the path to some area near the specified node is checked. This reduces the number of nodes
in the tree under construction. The developed algorithm is first compared with the traditional algorithm
of fast-growing random trees. The comparison is made by the trajectory calculation time, the
amount of memory required, the path length and the percentage of situations in which the trajectory
to the target point was successfully found. Next, the developed algorithm is compared with the planning
algorithms of other classes. The study uses representative samples of numerical experiments and
various environments that differ in the density of obstacles and the presence of mazes. A study of
planning algorithms using the results of experiments on a ground-based wheeled robot is also being
conducted. Based on the results of numerical and real experiments, conclusions are drawn about the
advantages and disadvantages of the developed algorithm of motion planning and the feasibility of its
application in various environments. -
EVOLUTIONARY POPULATION METHOD FOR SOLVING THE TRANSPORT PROBLEM
B.К. Lebedev, О.B. Lebedev, Е.О. Lebedevа2022-11-01Abstract ▼The paper considers an evolutionary population method for solving a transport problem
based on the metaheuristics of crystallization of a placer of alternatives. We study a closed (or
balanced) model of the transport problem: the amount of cargo from suppliers is equal to the total
amount of needs at destinations. The goal of optimization is to minimize the cost (achieving a minimum
of transportation costs) or distances and the criterion of time (a minimum of time is spent on
transportation). The metaheuristics of the crystallization of a placer of alternatives is based on a
strategy based on remembering and repeating past successes. The strategy emphasizes «collective
memory», which refers to any kind of information that reflects the past history of development and
is stored independently of individuals. An ordered sequence Dk of routes is considered as a code
for solving the transport problem. The objects are routes, the alternatives are the set of positions P
in the list, where np is the number of positions in the list Dk. The set of objects Dk corresponds to
the set of all routes. The set of alternative states P of the object corresponds to the set of alternative
options for placing the object in the list Dk. The operation of the population evolutionary algorithm
for the crystallization of a placer of alternatives is based on a collective evolutionary
memory called a placer of alternatives. A scattering of solution alternatives is a data structure
used as a collective evolutionary memory that carries information about the solution, including
information about the realized alternatives of agents in this solution and about the usefulness of
the solution. A constructive algorithm for the formation of a reference plan by decoding the list Dk
has been developed. At each step t, the problem of choosing the next route in the sequence Dk and
determining the amount of cargo transported from the point of departure Ai to the point of destination
Bj along this route is solved. The developed algorithm is population-based, implementing the
strategy of random directed search. Each agent is a code for some solution of the transport problem.
At the first stage of each iteration l, a constructive algorithm based on the integral placer of
alternatives generates nk decision codes Dk. The formation of each decision code Dk is performed
sequentially in steps by sequentially selecting an object and position. For the constructed solution
code Dk, the solution estimate ξk and the utility estimate δk are calculated. An individual scattering
of alternatives Rk is formed and a transition to the construction of the next solution code is formed.
At the second stage of the iteration, the integral placer of alternatives formed at previous iterations
from l to (l-1) is summed with all individual placers of alternatives formed at iteration l.
At the third stage of iteration l, all integral utility estimates r*
αβ of the integral placer of alternatives
R*(l) are reduced by δ*. The algorithm for solving the transport problem was implemented in
C++ in the Windows environment. Comparison of the values of the criterion, on test examples,
with a known optimum showed that in 90% of the examples the solution obtained was optimal, in
2% of the examples the solutions were 5% worse, and in 8% of the examples the solutions differed
by less than 2%. The time complexity of the algorithm, obtained experimentally, lies within O(n2). -
THE ADJACENCY MATRIX RECONSTRUCTION ALGORITHM FOR CAUSAL GRAPH MODELS IN THE ABSENCE OF OBSERVABLE VARIABLES
A. N. Tselykh , V.S. Vasilev, L. A. Tselykh2021-11-14Abstract ▼The paper deals with the problem of modeling complex systems in the absence of observable
variables. To solve this problem, it is proposed to use causal graph models. The class of causal
models considered here is defined as non-stochastic causal models with unobservable variables.
These models are presented in the form of a directed graph, created on the basis of human mental
representations. In this case, on the arcs, causality is expressed in the form of some marks with a
sign that determines the direction of change in the state of the system. The considered causal models
include heterogeneous, complex and qualitative types of variables that illustrate the nonnumerical
nature of nodes and links and, as a consequence, the absence and impossibility of obtaining
time series data. In the absence of observable variables and the impossibility of conducting
experiments, the problem of reconstructing the adjacency matrix of the causal graph model becomes
much more complicated. It is required to obtain a model with a certain spectral decomposition
that implements the main function of the modeled system. Based on this concept, a new method
for reconstructing the adjacency matrix is proposed, implemented on the basis of the corresponding
causal propagation matrix or transmission matrix. The idea is to use combinatorial optimization
based on spectral graph theory to generate data from a qualitative non-stochastic causal
model and reconstruct an adjacency matrix using that data. In this case, the eigenvectors are
identified as key objectives of the matrix reconstruction process, which postulates a fundamental
approach based on the spectral properties of the graph. The results of computational experiments
on solving the problem of reconstructing the adjacency matrix for causal graph models in the absence
of observable variables using the developed algorithm have shown that the algorithm effectively
reconstructs matrices from the given parameters with admissible similarity indices. The
convergence of the approximation to the solution of the matrix reconstruction algorithm is proved
no slower than with the speed of a geometric progression. From a technical point of view, the
advantage of the algorithm is the implementation of a tool for automatic adjustment of the regularization
parameter, suitable for users without prior mathematical knowledge. -
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.
-
MODERN APPROACHES TO FACE RECOGNITION IN LOW-LIGHT CONDITIONS: A REVIEW AND THE CONCEPT OF A HYBRID END-TO-END ARCHITECTURE
D. А. Morozov , V.V. Gilka , А. S. Kuznetsova113-1332026-07-07Abstract ▼The article addresses the problem of reliable face recognition in critical areas such as video surveillance and biometric authentication under low-light conditions. Existing approaches typically separate the tasks of image enhancement and face identification, which leads to error accumulation and loss of informative features. The aim of this work is to overcome this limitation by developing and theoretically substantiating a hybrid end-to-end architecture in which image enhancement and face recognition are solved jointly. The study provides a systematic review of modern methods, including classical algorithms (such as histogram equalization and noise suppression) and advanced deep neural networks (including EnlightenGAN, Zero-DCE, ArcFace, and RetinaFace). The main contribution is the integration of generative and identification modules into a single computational graph. The key result of the study is the demonstration that joint optimization of all processing stages within a unified model, unlike fragmented solutions, fundamentally changes the approach to the problem. Theoretical analysis and comparative evaluation of existing concepts show that the proposed architecture ensures a more efficient gradient flow during training, leading to the formation of higher-quality and noise-robust identity features. It is shown that this approach prevents error accumulation between stages and minimizes information loss. The novelty of the work lies in the holistic, end-to-end view of the face recognition problem under low-light conditions. The practical significance is confirmed by the applicability of the architecture in real systems, where its implementation can potentially improve reliability and processing speed by combining heterogeneous tasks into a single optimizable framework.
The article addresses the problem of reliable face recognition in critical areas such as video surveillance and biometric authentication under low-light conditions. Existing approaches typically separate the tasks of image enhancement and face identification, which leads to error accumulation and loss of informative features. The aim of this work is to overcome this limitation by developing and theoretically substantiating a hybrid end-to-end architecture in which image enhancement and face recognition are solved jointly. The study provides a systematic review of modern methods, including classical algorithms (such as histogram equalization and noise suppression) and advanced deep neural networks (including EnlightenGAN, Zero-DCE, ArcFace, and RetinaFace). The main contribution is the integration of generative and identification modules into a single computational graph. The key result of the study is the demonstration that joint optimization of all processing stages within a unified model, unlike fragmented solutions, fundamentally changes the approach to the problem. Theoretical analysis and comparative evaluation of existing concepts show that the proposed architecture ensures a more efficient gradient flow during training, leading to the formation of higher-quality and noise-robust identity features. It is shown that this approach prevents error accumulation between stages and minimizes information loss. The novelty of the work lies in the holistic, end-to-end view of the face recognition problem under low-light conditions. The practical significance is confirmed by the applicability of the architecture in real systems, where its implementation can potentially improve reliability and processing speed by combining heterogeneous tasks into a single optimizable framework.
-
IMPROVING INTERFERENCE IMMUNITY OF GROUND-TO-AIR RADIO LINKS BY ADAPTIVE ROUTE MODIFICATION OF A UAV RELAY BASED ON INTERFERENCE ENVIRONMENT ANALYSIS
А. А. Shmidt , V.R. Bikbulatov , D.N. Polyakov , А.А. Tkhakakhov2026-04-29Abstract ▼The relevance of the work is driven by the increasing intensity of electronic countermeasures in the tactical command echelon, where traditional relay communication methods with fixed routes fail to ensure the required interference immunity and signal security. The objective of this study is to develop a method for adaptive route control of a UAV relay based on continuous monitoring of the electromagnetic environment in order to improve interference immunity of ground-to-air radio links. The primary method employed is multi-criteria flight path optimization with adaptive weighting coefficients, simultaneously considering three criteria: minimization of interference levels at the relay operating frequencies, ensuring radio accessibility with network correspondents, and reduction of detection probability by enemy electronic warfare assets. To reconstruct the spatial interference pattern from a limited set of onboard measurements, several interpolation and extrapolation methods are examined: inverse distance weighting, radial basis function interpolation, and statistical extrapolation based on a spatial correlation function. A cyclic operational procedure for the adaptive routing system is developed, comprising data collection, construction of a three-dimensional interference map, prediction of its evolution, calculation of the optimal trajectory, and monitoring of the maneuver outcome. Simulation results show that the proposed method increases the signal-to-interference ratio by 1.5–2 dB on average and up to 8 dB in the worst-case scenario compared to fixed-route flight. The practical significance lies in the possibility of implementing the proposed method using existing UAV platforms and radio-electronic equipment without fundamentally new technical solutions.
-
A BIOINSPIRED APPROACH TO SOLVING THE PROBLEM OF 3D PACKAGING
V.I. Danilchenko , V.V. Bova , М. М. Semenova , S.V. Ignateva , М. B. Shayliev2026-02-27Abstract ▼This article examines one of the most important combinatorial optimization problems – three-dimensional packaging. Optimizing three-dimensional packaging reduces costs and improves logistics efficiency, making it relevant for industry. This paper analyzes classical approaches such as greedy algorithms and dynamic programming, as well as widely used methods, including evolutionary algorithms and local search. An analysis of existing methods, including greedy search, dynamic programming, evolutionary algorithms, and local search, revealed their key characteristics and identified suitable areas of application. In the context of this analysis, an overview of the key methods that dominated during certain historical periods is presented. The analysis includes consideration of the application conditions of various methods, their effectiveness for specific types of problems, as well as their advantages and limitations.
A multi-level search algorithm is presented that combines the advantages of traditional and modern optimization methods. This multi-level algorithm improves the accuracy of the packaging problem solution through dynamic parameter adjustment. A software package for solving the three-dimensional packaging optimization problem using bioinspired algorithms has been developed. A computational experiment was conducted on test examples (benchmarks). The packing quality obtained using the developed combined bioinspired algorithm is, on average, 7% higher than the packing results obtained using known algorithms, while the solution time is 7% to 25% shorter, demonstrating the effectiveness of the proposed approach. A series of tests and experiments allowed us to refine theoretical estimates of the time complexity of packing algorithms. In the best case, the time complexity of the algorithms is O(n²), and in the worst case, O(n³). -
DEVELOPMENT AND IMPLEMENTATION OF A CORPORATE INFORMATION SYSTEM AT THE AvtoVAZ INDUSTRIAL ENTERPRISE
D.Y. Zorkin , А.А. Bognyukov , Т. Е. Kozhanova2026-02-27Abstract ▼In the context of global industrial digitalization, the development and implementation of corporate information systems (CIS) have become strategically critical for enhancing operational efficiency and competitiveness of enterprises. This study examines the integration case of the ERP system "1C: Enterprise Management" at the AvtoVAZ plant – a key player in the Russian automotive industry. The research aimed to optimize management and production processes through the automation of planning, resource accounting, and coordination of cross-functional interactions. The methodological framework combined analytical, graphical, and comparative approaches, as well as practical testing of solutions in the "1C" software environment. The focus was on designing algorithms for managing production cycles, forming resource specifications, and configuring planning scenarios. The study developed demand forecasting models, analyzed production capacities, and balanced output based on model prioritization (Lada Granta, Vesta, Largus). The system implementation reduced order processing time by 30%, minimized warehouse downtime by 18–22% through synchronized logistics schedules, and improved quality control accuracy via integrated diagnostic tools (CAN-bus, spectrophotometry). Special emphasis was placed on overcoming institutional and technological barriers, including modernizing outdated planning methods, training employees in ERP interfaces, and deploying hybrid cloud solutions to ensure system scalability. The practical significance of the research was confirmed by achieving resource allocation transparency, reducing operational costs, and forming an adaptive production strategy aligned with market dynamics. The results demonstrate that CIS implementation not only optimizes current business processes but also lays the foundation for sustainable enterprise development in the digital transformation era. The acquired experience can be extrapolated to other engineering and industrial enterprises facing challenges in management automation and data integration under competitive pressure. Future research prospects involve analyzing the long-term effects of ERP system adoption, including their impact on innovation potential and supply chain ecosystems.
