Harvester open publications of NAS Ukraine
Harvester open publications of NAS UkraineNot a member yet
2350 research outputs found
Sort by
Method of managing the execution of tasks of a multithreaded program according to a given dependency graph
No full textThis article examines the effectiveness of pre-training generative model based on a visual transformer and subsequent fine tuning for image classification tasks. The main problem of the study is the poor training efficiency of the visual transformer on a limited amount of data. It is possible to improve the accuracy of the image classification model by using transfer learning of the knowledge obtained during the previous training of the generative model on the same data. A subset of the standard Imagenet dataset - Tiny Imagenet was used to test the hypothesis. It contains 200 categories of around 500 images each. The size of each image is 64x64 pixels. For pre-training the generative model, patches are used to mask image segments. The training of restoring masked image pixels forces the model to pay attention to the context around the removed part, as well as to general visual patterns. This leads to a better understanding of visual information by the model as a whole and helps with further fine tuning of the model for the classification task. As a result of a series of experiments, it was possible to achieve an improvement in the accuracy of image classification from 40% to 44.7%, and an analysis of the effect of the overall degree of masking and patch size on it is given. Additionally, impact of different sizes of patches (2x2, 4x4, 8x8 pixels) and different percentages of masking (20/40/60 percent) of the input image were investigated in the paper.Prombles in programming 2024; 2-3: 247-252
Video-based visualization of debugging process
Get PDFOne of the current trends in education is microlearning, which involves the use of short videos in the learning process. Microlearning has a number of advantages, including the fact that this approach is more student-centered, aims to increase the level of knowledge assimilation, requires less time for learning, and allows learning anytime and anywhere. In previous works, the authors have developed a constructive production model and corresponding software for tracking programmer's actions during the preparation of program text and debugging in the Visual Studio development environment. This article presents an extension of these software tools. Based on the collected information in the log files about the program debugging processes, their visualization is performed, which reproduces the sequence of actions during the original debugging process. The goal is to increase the efficiency and effectiveness of programming education. The video-based visualization demonstrates the programmer's work on preparing and correcting the code during debugging and is synchronized with the time stamps in the log files only during periods of activity. Comments are overlaid on the video, providing explanations and suggestions for improving the debugging process. Comments help to understand the rationale for specific actions taken during debugging and provide guidance on how to improve processes or use alternative approaches. The benefit of visualization for the teacher is the ability to: analyze the debugging process of a particular student, identify typical mistakes of a particular group of students, adjust the teaching process accordingly, and provide targeted assistance in improving debugging skills. Benefits for the student: the ability to analyze your own work, develop critical thinking on how to improve it, and receive timely assistance from the teacher.Prombles in programming 2024; 2-3: 426-433
Development of the intelligent control system of an unmanned car
Get PDFThis study delves into creating an intelligent control system for self-driving vehicles, utilizing cutting-edge machine learning methods. Central to our approach is the NeuroEvolution of Augmenting Topologies (NEAT) algorithm, implemented in the Python programming language. NEAT plays a pivotal role in refining artificial neural networks, enabling autonomous cars to navigate diverse road conditions independently. Through rigorous experimentation, we demonstrate NEAT's capability to automate self-driving operations, ensuring adaptability to various driving scenarios. The result of the research is the development of a complex system proficient in autonomously navigating a variety of race tracks. NEAT's dynamic neural network structures help the vehicle learn quickly.The Python language is quite convenient for implementing such tasks thanks to a large number of libraries. Integration with Pygame equips the system with essential tools for graphics rendering and interaction. Iterative cycles of training and refinement significantly enhance the system's performance and adaptability. Neural networks adeptly learn to navigate tracks, maintain optimal speeds, avoid collisions, and tackle diverse racing challenges. This project demonstrates NEAT's capability, alongside Python and Pygame integration, in crafting intelligent control systems for self-driving cars. This holds promise for further development in autonomous driving technology, aiming to handle more intricate scenarios and seamlessly integrate with real-world hardware. In essence, the successful deployment of an intelligent control system for unmanned vehicles based on NEAT demonstrates the efficacy of evolutionary algorithms in tackling complex control problems. This sets the stage for further research and refinement in unmanned driving, fostering the development of safer and more efficient transportation systems.Prombles in programming 2024; 2-3: 375-38
A model of centralized supply chains with independent behavior of separate nodes
Get PDFThis paper proposes a model of supply chains combining a centralized structure with independent behavior of individual nodes. The peculiarity of this model is that it finds application in the modeling of decentralized big data systems, which have become widespread recently. To build the model, existing architectures and approaches, in particular from the theory of automatic control, were considered. These approaches made possible to choose the most appropriate approach to represent the big data network dynamics and, accordingly, its behavior in time. In the proposed model, this is achieved by using a centralized approach to the construction of network architecture and modeling the behavior of network nodes and individual chains with a model predictive control. As part of the study, the problem of the three-dimensional forecasting horizon is posed, which consists in the need to describe the dynamics in three coordinates, which are responsible for the spread of the solution in depth, width and time, which clearly affects the complexity and the possibility of its solution in an acceptable time. In order to solve this problem, we propose to split the model into separate coordinates, which allows solving the spatio-temporal representation of nodes and, accordingly, the state space model by separate systems of dynamics equations - in space and time. To test the model, an experimental implementation was created, which implements the tasks of modeling network dynamics of the model with the involvement of neuro-optimal regulators, based on Pontryagin’s principal of maximum - for temporal dynamics and a predictive control model for spatial network dynamics, respectively. As a result of the experimental tests of the model, an assessment of the adequacy of the model was given and general recommendations for the development of supply chain models were given, as well as possible potential advantages of using neuro-optimal regulators compared to the predictive control model were indicated.Prombles in programming 2024; 2-3: 305-31
Automated parallelization of a program for modeling intraparticle diffusion and adsorption in heterogeneous nanoporous media
No full textHeterogeneous media consisting of thin layers of particles of forked porous structure with different physical-chemical properties are widely used in science-intensive technologies and priority sectors of industry, medicine, ecology, etc. Such layers are distributed systems of pores consisting of two main spaces: micro- and nanopores of particles and macropores and cavities between particles. Mass transfer in the system of heterogeneous media causes two types of mass transfer: diffusion in macropores, owing to interparticle space, and diffusion in the system of micro- and nanopores inside particles of the heterogeneous medium. Intraparticle space has a higher level of adsorptive capacity, and at the same time, has a lower velocity of diffusion intrusion in comparison with interparticle space. In modeling concentra- tion and gradient fields for various diffusible components, an important scientific problem is the identification of kinetic parameters of a transfer, predetermining mass transfer velocity on macro- and micro levels, and also equilibrium conditions. The results of designing and parallelization of a program implementing a Crank-Nicolson scheme using algebra-algorithmic specifications represented in a natural- linguistic form are given. The tools for automated design, synthesis and auto-tuning of programs were applied that provided the translation of algebra-algorithmic schemes into source code in a target programming language and its tuning for execution environment to increase the program performance. Numerical distributions of values of diffusion coefficients for intraparticle transfer along coordinate of medium thickness for various time snapshots were obtained. Based on the results of the identification, the models were checked for adequacy and numerical modeling and analysis of concentration and gradient fields of mass transfer were carried out. The experiment results of auto- tuning the software implementation demonstrated high multiprocessor speedup on test data input.Prombles in programming 2022; 3-4: 59-6
Research on methods of building an interactive map in a web application
Get PDFThe purpose of this article is to formulate a list of rules for creating an efficient and convenient informational system for vicinity and building navigation and to create a web-app with this functionality. To accomplish this, a web-app user experience, navigation data representation options, their user perception and influence on their emotions were analyzed. Also, a list of recommendations for data representation and storing inside the app alongside with ways of fetching them was composed.General navigational application development practices were described in this work, a test version of an application was created to harvest users’ reviews and feedback and a "University interactive map" web-application for classroom, teacher rooms, departments, laboratories information displaying and navigating buildings and vicinities was created.Prombles in programming 2022; 3-4: 327-33
Friend-or-Foe Recognition Algorithm Development for the Corresponding Software Building
Get PDFThe year 2022 showed an urgent need to improve the existing systems for recognizing objects in the aerial space, which is caused by the significant increase in the number of technical means (especially unmanned aerial vehicles) on the battlefield. Such a sharp increase in the number of objects that simultaneously take part in combat operations in the air requires the improvement of military object recognition systems, both qualitatively and quantitatively. This requires the development of appropriate new generation Friend-or-Foe algorithms for the objects’ recognition.The main requirements for recognition systems of aerial objects of civil application were determined. They includes: maximum com- patibility; support for a large number of objects; outdated recognition complexes support; support for alternative ways of recognition; support for alternative data entry methods; determining the coordinates of aerial objects in an emergency situation.Friend-or-foe recognition systems for military applications are also considered. In contrast to civilian systems, the following basic requirements have been identified for them: 1) Maximum speed of the recognition process. 2) Protection against false positives. 3) Protection against legitimate aerial object imitation. 4) Support for a large number of objects. 5) Protection against cases of loss of a legitimate aerial object. 6) Rotation of the secret part. 7) Protection against false-negative results to prevent friendly fire. 8) Protection against man-in-the-middle attacks. 9) Flexible integration with the NATO block recognition system. 10) Availability of opportunities for purely domestic production and support of the object recognition system. 11) Protection against electronic warfare means. 12) Support for several recognition modes. 13) Automatic blocking of the launch of ground-to-air and air-to-air weapons against objects that confirm their legitimacy by a correct response to a request. 14) Determining the coordinates of aerial objects in an emergency.Based on the formulated requirements, a new friend-or-foe algorithm for the state identification system for military use is proposed, built based on the state standards, and taking into account the features of its software implementation in order to increase speed. Its implementation will ensure sufficient scalability, stability, reliability, and multi-level recognition.Prombles in programming 2022; 3-4: 387-39
Creation and test of applied software of network of wireless sensors for agriculture
Get PDFThe article describes applied software of units of such complex hardware-software system, as plants` state monitoring system for application in agriculture and ecological monitoring. The mentioned system consists of data acquisition system in the form of wireless sensor network and adaptive part in the form of decision-making support system. The authors described main applied software of au- tonomous nodes of wireless sensor network and implementation of some program functions of decision-making support system. Wire- less sensor network includes many autonomous wireless sensors, so the main criteria during applied software creation was assuring the energy efficiency of operation of autonomous measuring nodes and network coordinator, and correct interaction of nodes within all network. As it is very difficult to perform testing of applied software of wireless nodes individually in field conditions, it was tested the network cluster, including hardware and software as a whole, in conditions like to applied task. The main parameters, which define the correctness of applied software operation, were estimated. These parameters include, for example, time of network selforganization, distance and quality of stable communication, time of autonomous operation of wireless nodes without charging batteries and so on. To create applied software for the decision-making support system, first of all, methods of plants` state diagnosing and estimating the factors, which influence the plant state, were developed. For this, the field experiments were conducted to determine sufficient dose of herbicide application and estimate the soil moisture using the chlorophyll fluorescence induction method. For processing measured data, several methods of machine learning were used, including neural network approach. Application of machine learning methods made it possible, on the base of acquired data, to make early diagnostics of influence of stress factors on the plant even before the appearance of visual manifestations of such negative influence and determine the decrease of soil moisture through the diagnostics of plant itself, and inform the user about this.Prombles in programming 2022; 3-4: 425-43
A GPU-based singular value decomposition algorithm
No full textIn this research paper we present an implementation of a singular value decomposition algorithm designed specifically for the graphics processing unit. It consists of two parts: orthogonal matrix decomposition and matrix diagonalization. Presented an implementation of bidiagonalization algorithm where we calculate the main bidiagonal matrix and two orthogonal multipliers using a series of House- holder transformations, as well as diagonalization algorithm with the help of Givens rotation matrices. Bothe these parts are implemented in jCUDA environment. Experiments have been conducted, the results of which have been thoroughly investigated on the matter of time consumption and calculations error. We’ve also compared our implementation with alternatives both on central and graphic processors.Prombles in programming 2023; 1: 30-3
Development and implementation of the management system of the economic and property complex of the National Academy of Sciences of Ukraine
No full textIn modern conditions, the issue of building an effective management system for the economic and property complex of the National Academy of Sciences of Ukraine becomes an urgent issue. This can be done on the basis of the creation and implementation of computer information and analytical systems, in particular, the Digital Real Estate Management System of the National Academy of Sciences of Ukraine (hereinafter - CS PROPERTY). CS MAYNO is intended for management of accounting, storage and use of reliable and updated information on real estate, which is recorded on the balance sheet of institutions, organizations, enterprises of the National Academy of Sciences of Ukraine, and management of lease and use contracts. The system must be developed using a modern software platform without using the software of the aggressor country.In this paper, a comprehensive analysis of the subject area is carried out and ways of automating the activities of the National Academy of Sciences of Ukraine in this area are described.Prombles in programming 2023; 2: 24-3