2024
Постійне посилання на фонд
Переглянути
Перегляд 2024 за Назва
Зараз показуємо 1 - 20 з 43
Результатів на сторінці
Налаштування сортування
Документ Відкритий доступ Analisysis of the electromagnetic processes in converter with eleven zone voltage regulation(КПІ ім. Ігоря Сікорського, 2024) Mykhailenko, V.; Trotsenko, Y.; Chunyak, J.The article focuses on the analysis of electromagnetic processes in the electric circuits with semiconductor switches. The mathematical model of the converter with fourteenzone regulation of output voltage has been developed to analyse electromagnetic processes in semiconductor converters with pulse-width regulation. The graphs representing electromagnetic processes in electric circuits are given.Документ Відкритий доступ Application of embeddings for multi-class classification with optional extendability(КПІ ім. Ігоря Сікорського, 2024) Smilianets, F.This study investigates the feasibility of an expandable image classification method, utilizing a convolutional neural network to generate embeddings for use with simpler machine learning algorithms. The possibility of utilizing this approach to add new classes by additional training without modifying the topology of the vectorization network was shown on two datasets: MNIST and Fashion-MNIST. The findings indicate that this approach can reduce retraining time and complexity, particularly for more complex image classification tasks, and also offers additional capabilities such as similarity search in vector databases. However, for simpler tasks, conventional classification networks remain more time-efficient.Документ Відкритий доступ Application of transfer learning for enhanced pulmonary disease detection via ct image embeddings(КПІ ім. Ігоря Сікорського, 2024) Smilianets, F.This paper presents a method for сomputed tomography imaging analysis for disease diagnosis, extending and fine-tuning a previously trained network to generate embedding vectors. A KNeighborsClassifier trained on produced embeddings achieved an accuracy of 0.987.Документ Відкритий доступ Automated healthcare systems’ review(КПІ ім. Ігоря Сікорського, 2024) Kravets, М.; Batrak, Y.; Tsopa, N.; Markin, M.An automated healthcare system is a high-tech integrated approach to monitoring individuals' health and providing appropriate recommendations. The goal is to review and analyze existing approaches for measuring healthcare parameters. Various existing applications are analyzed to select the best implementations and correct their shortcomings.Документ Відкритий доступ CNN for solving Computer vision tasks(КПІ ім. Ігоря Сікорського, 2024) Kovalchuk, R.; Polshakova, O.Аbstract: The paper explores fundamental and current methods addressing computer vision tasks, particularly in classification, segmentation, and image processing implemented in computer vision (CV) systems. It is highlighted that the application of deep learning significantly enhances the speed and accuracy in processing large datasets within CV systems. It is noted that artificial vision systems are proficient in object detection, recognizing serial numbers, and detecting surface defects. Emphasis is placed on the significance of integrating diverse methods for effectively addressing CV tasks. Directions for further research are proposed, such as the application of CV systems in resource-constrained environments and enhancing real-time processing efficiency.Документ Відкритий доступ Convolutional neural network for dog breed recognition system(КПІ ім. Ігоря Сікорського, 2024) Khotin, K.; Shymkovych, V.; Kravets, P.; Novatsky, A.; Shymkovych, L.In this article a dataset with data augmentation for neural network training and convolutional neural network model for dog breed recognition system has been developed. Neural network model architecture using transfer learning to improve classification results was developed. Neural network based on the MobileNetV3-Large architecture. The structure of the dataset has analyzed and decision to use different methods for normalizes data. A large dataset containing 70 distinct categories of dog breeds was collected and balanced through the use of data augmentation techniques. Data augmentation enabled the reduction of the disparity between the minimum and maximum number of instances by eliminating redundant images and adding essential ones. The developed model was tested and the results were demonstrated. The final accuracy of the model is 96%. The result model implement in dog breed recognition system, which is based on mobile platform. The implemented application produces functionality to interact with the resulting model such as real-time process of identifying a dog's breed or from device's gallery. Further improvement the performance of the model classification quality can be achieved by expending the initial dataset or by applying other optimization methods and adjust the learning rate.Документ Відкритий доступ Development of subject-oriented innovative software products based on artificial intelligence systems(КПІ ім. Ігоря Сікорського, 2024) Stenin, A.; Pasko, V.; Lisovychenko, O.; Polshakova, O.; Soldatov, V.This article analyzes existing models of the development process of innovative software products for a specific subject area. Taking into account the multialternative nature of solutions at each stage of innovative software products development and compliance with constantly changing requirements, this article proposes an information and logical model of innovative software products development management process based on artificial intelligence systems. The essence of the proposed approach is that the development of an innovative software products is interpreted as an information object that changes in content and structure in the process of its creation. The presence of uncertainty and a large amount of information requires the use of intelligent decision support systems at each stage of ISP development.Документ Відкритий доступ Effect of rendering space onthe performance of the ray marching method(КПІ ім. Ігоря Сікорського, 2024) Danyliuk, Y.; Zhurakovska, O.This study investigates the efficiency of rendering implicit surfaces using two distinct approaches: screen rendering space and bounding volumes rendering space. Through comprehensive analysis, visualization of rendering differences, and performance testing, we evaluate the effectiveness of each approach. Our findings demonstrate significant advantages offered by bounding volumes rendering space, including efficient elimination of empty areas, reduction of ray marching iterations, and enhanced CPU and GPU resource utilization. Notably, bounding volumes rendering space achieves an 87% increase in average framerate compared to screen rendering space, achieving 102.71 FPS versus 54.82 FPS, respectively. Our results underscore the promise of bounding volumes rendering space as a potent approach for improving rendering efficiency in ray marching-based implicit surface visualization. We advocate for its integration into rendering pipelines and suggest avenues for future research to maximize its benefits and impact on computer graphics and visualization.Документ Відкритий доступ Energy and water consumption monitoring system for co-owners of an apartment building(КПІ ім. Ігоря Сікорського, 2024) Storchun, V.; Mamedova, K.The problem of collecting, analyzing and distributing data collected from energy and water sensors between co-owners of an apartment building is considered. A comprehensive solution is proposed, including automated monitoring and notification system, which allows to increase the efficiency of expenses management and improve reaction time in an emergency.Документ Відкритий доступ Evaluation of unstructured resumes using the Word2Vec model(КПІ ім. Ігоря Сікорського, 2024) Martseniuk, K.; Deveciogullari, A.The article addresses the problem of evaluating candidates' resumes for job vacancies using various natural language processing (NLP) methods. Traditional text processing algorithms were analyzed, and a critical drawback of these methods was identified—their inability to account for semantic relationships between words, which is particularly important in the context of resume evaluation. The BERT model was also considered, but it was dismissed due to its high computational complexity and excessive functionality for this task. The primary choice for evaluating resume relevance was the Word2Vec method, which accounts for semantic relationships between words, contributing to greater objectivity and accuracy in the evaluation process. The study results confirm the effectiveness of using Word2Vec compared to other methods in the context of resume analysis.Документ Відкритий доступ Intelligent control system with reinforcement learning for solving video game tasks(КПІ ім. Ігоря Сікорського, 2024) Osypenko, M.; Shymkovych, V.; Kravets, P.; Novatsky, A.; Shymkovych, L.This paper describes the development of a way to represent the state and build appropriate deep learning models to effectively solve reinforcement learning video game tasks. It has been demonstrated in the Battle City video game environment that careful design of the state functions can produce much better results without changes to the reinforcement learning algorithm, significantly speed up learning, and enable the agent to generalize and solve previously unknown levels. The agent was trained for 200 million epochs. Further training did not improve results. Final results reach 75% win rate in the first level of Battle City. In most of the 25% of games lost, the agent fails because it chooses the wrong path to pursue an enemy that is closer to the base and therefore slower. The reason for this is the limitation of cartographic information. To further improve performance and possibly achieve 100% win rate, it is recommended to find a way to effectively include full information about walls and other map objects. The developed method can be used to improve performance in real applications.Документ Відкритий доступ Models for analyzing and forecasting share prices on the stock exchange(КПІ ім. Ігоря Сікорського, 2024) Piznak, R.; Likhouzova, T.The work is devoted to the analysis and forecasting of share prices for four leading technology companies: Nvidia, Apple, Google, and Netflix. These companies are leaders in their fields and have a significant impact on the global economy. The goal is to study the dependencies affecting the share prices of companies, as well as to develop models for forecasting future trends. In the work, a thorough analysis of historical data on company share prices and their macroeconomic indicators was carried out. The study was based on the fundamental concepts of economic science. The study results are expected to provide a deeper understanding of the prospects of these companies.Документ Відкритий доступ Models for analyzing the complexity of english words in the text on the scale from A1 to C2(КПІ ім. Ігоря Сікорського, 2024) Bielikov, M.; Likhouzova, T.; Oliinyk, Y.At the current stage of globalization, English plays a key role as the language of international communication. This leads to the fact that more and more people become its carriers at various levels. The work is devoted to the analysis of English words on the scale from A1 to C2, which corresponds to the lowest and highest levels of proficiency according to the CEFR standards. A model that predicts the difficulty of words in a text can be used to improve the educational process. For example, it is possible to find a list of likely unknown and difficult words for the end user in any text depending on his level of English language proficiency. This approach will facilitate the language learning process by providing a personalized list of words to focus on. Also, the model can be useful for analyzing the complexity of texts depending on the number of words of each level of complexity in them. This can help teachers prepare materials that match the level of knowledge of their students, as well as identify words that may be difficult for them to understand. An application in the Python programming language is proposed, which receives a sample of data from the created storage, displays them graphically, performs intellectual analysis, trains and compares models according to accuracy, precision, recall and f1-score metrics. For data analysis and prediction of the level of complexity of English words, the following models were used: PchipInterpolator, logarithmic model, Gradient Boosting, Random Forest and XGB.Документ Відкритий доступ Models for researching prospects for the development of the aviation industry(КПІ ім. Ігоря Сікорського, 2024) Tarasonok, D.; Likhouzova, T.Abstract: Considered the problem of prospects for the development of the aviation industry in the world. It is proposed to use machine learning models and technologies for clustering countries by the number of air flights and forecasting the production of aircraft in the future. Several different quality measures were used to evaluate the effectiveness of the proposed models, which reflect the needs of each task in various ways.Документ Відкритий доступ Modern technologies for hiding people's faces using object tracking based on YOLOv5 and deepsort(КПІ ім. Ігоря Сікорського, 2024) Shchur, А.; Polshakova, O.The object of study is a system for automated blurring of human faces in video. This article provides a detailed overview of modern technologies and principles of tracking objects in video with assigning them unique elements. Since most video editors still leave most of the work to the user, it was decided to optimize this process. The aim of this work is to reduce the time spent on the process of hiding human faces in video files. To achieve this goal, it is proposed to use a modern detector - the YOLO convolutional neural network and the DeepSORT object tracking algorithm, which uses classical approaches to filtering input data and predicting the position of an object in space, as well as a modern neural network capable of distinguishing between people's faces. As a result of this work, among free analogues on the Internet, the acceleration of face blurring was achieved up to 20%, which is a pretty good result.Документ Відкритий доступ Operational control of city engineering networks based on artificial intelligence systems(КПІ ім. Ігоря Сікорського, 2024) Stenin, A.; Pasko, V.; Soldatova, M.; Drozdovich I.This article proposes a situational approach to the construction of intelligent decision support systems for the operational management of city utility networks. This approach is based on the use of a directed weighted graph of fuzzy situations. The situational fuzzy control algorithm presented in the article is based on the expert method of forming and evaluating alternatives to management decisions and the structural generality of the oriented graph of the fuzzy situational network.Документ Відкритий доступ Optimization of electricity costs in residential heat supply systems(КПІ ім. Ігоря Сікорського, 2024) Stenin, A.; Pasko, V.; Soldatova, M.; Drozdovich, I.Currently, there is a large number of objects related to recirculation of material flows, mixing of reagents of liquid and gaseous media, etc. Among them are heat supply systems of residential buildings. Their dynamics are described by differential equations with a deviating argument. Optimizing the operation of such objects using known methods is quite difficult. This paper proposes a dummy variable method for reducing the original system of differential equations with a deviating argument to a system of ordinary differential equations and transfer equations for which known optimization methods can be applied. The method is generalized to systems with a non-stationary deviating argument and to systems with several processes having an aftereffect. The practical use of the dummy variable method is shown on the example of solving the problem of optimal control of the heat exchange process in a private residential building.Документ Відкритий доступ Soft skill in it students training(КПІ ім. Ігоря Сікорського, 2024) Gashimova, M.; Segol, R.; Batrak, Y.; Tsopa, N.; Murakhovskyy, S.; Harmash, O.The article is devoted to developing soft skills for IT professionals in the process of remote learning through the introduction of team tasks using communication skills, report and presentation skills, teamwork, leadership skills, and activities under conditions of uncertainty. The experiment conducted within the framework of the study will allow us to scale the proposed methodology in the training of IT specialists in various universities. This article aims to introduce a methodology for teaching professional disciplines using the method for developing soft skills for future IT specialists. A method of teaching that helps students in a distance format, in addition to professional skills, also develops communication skills, presentation and report preparation skills, teamwork skills, emotional intelligence, and critical thinking.Документ Відкритий доступ Software evolution from system perspective(КПІ ім. Ігоря Сікорського, 2024) Linevych, O.; Lisovychenko, O.Abstract: Currently, software evolution takes an unpredictable amount of time. The purpose of the article is to describe the study of 7 popular and 1 new development method and their impact on the rate of evolution. The new method is the most effective, as it collects the most information on the connections between data and processes.Документ Відкритий доступ Study of the electromagnetic processes in converter with twenty four zoned regulations of the voltage(КПІ ім. Ігоря Сікорського, 2024) Mykhailenko, V.; Chunyak, J.; Prudnikov, M.Abstract: The article deals with electromagnetic processes in electric circuits with semiconductor switches. A mathematical model of a semiconductor converter with twenty four-zone output voltage control has been created so as to analyse electromagnetic processes in semiconductor converters with pulse-width control. The graphs showing electromagnetic processes in electric circuits are given.
- «
- 1 (current)
- 2
- 3
- »