Дисертації та автореферати
Постійне посилання на фонд
У фонді розміщено автореферати дисертацій та дисертації, захищені в КПІ ім. Ігоря Сікорського.
Переглянути
Перегляд Дисертації та автореферати за Назва
Зараз показуємо 1 - 20 з 1227
Результатів на сторінці
Налаштування сортування
Документ Відкритий доступ Cистема векторного керування асинхронним генератором з властивостями робастності до параметричних збурень(2021) Желінський, Микола Миколайович; Пересада, С. М.Документ Відкритий доступ Cистема векторного керування асинхронним генератором з властивостями робастності до параметричних збурень(КПІ ім. Ігоря Сікорського, 2021) Желінський, Микола МиколайовичДокумент Відкритий доступ Cистемно-структурний підхід до формування економічної стійкості машинобудівних підприємств(КПІ ім. Ігоря Сікорського, 2018) Кравченко, Марина ОлегівнаДокумент Відкритий доступ Cистемно-структурний підхід до формування економічної стійкості машинобудівних підприємств(2018) Кравченко, Марина Олегівна; Войтко, Сергій ВасильовичДокумент Відкритий доступ Cоціально-психологічні умови розвитку інноваційного потенціалу студентів закладу вищої технічної освіти(2021) Боковець, Ольга Ігорівна; Волянюк, Наталія ЮріївнаДокумент Відкритий доступ Cтратегії розвитку сервіс-орієнтованих систем у хмарному середовищі(2018) Петренко, Олексій Олексійович; Кисельов, Геннадій ДмитровичДокумент Відкритий доступ Cтратегії розвитку сервіс-орієнтованих систем у хмарному середовищі(КПІ ім. Ігоря Сікорського, 2018) Петренко, Олексій ОлексійовичДокумент Відкритий доступ Research and development of self-supervised visual feature learning based on neural networks(Igor Sikorsky Kyiv Polytechnic Institute, 2024) Xu Jiashu; Stirenko, SergiiXu Jiashu. Research and development of self-supervised visual feature learning based on neural networks. - Qualified scientific work on the rights of the manuscript. Dissertation for the degree of Doctor of Philosophy in the specialty 121 - Software Engineering and 12 - Information Technologies. - National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute", Kyiv, 2024. This Dissertation focuses on in-depth exploration into the design and development of self-supervised learning algorithms, which are a subset of unsupervised learning techniques that operate without the need for labeled datasets. These algorithms are particularly adept at pre-training models in an unsupervised manner, with the resultant models demonstrating performance on par with their supervised counterparts across a range of downstream applications. This method is particularly advantageous as it aims to mitigate the over-dependence on extensive data labeling that is typical within deep learning paradigms, thereby enhancing efficiency and practical utility in diverse real-world scenarios. The pertinence of selfsupervised learning algorithms is especially highlighted within the realm of medical image analysis. In this specialized field, the requisites for data annotation are not only laborious but also require a high degree of precision due to the critical nature of the data involved. The difficulty of obtaining accurate annotations is compounded by the scarcity of specialists capable of providing them, which in turn underscores the transformative potential of self-supervised learning approaches within this domain. In this dissertation, a cutting-edge self-supervised learning methodology is delineated, which employs the Mixup Feature as the reconstruction target within the pretext task. This pretext task is fundamentally designed to encapsulate visual representations by the prediction of Mixup features from masked image, utilizing these feature maps to extracting high-level semantic information. The dissertation delves into the validation of the Mixup Feature's role as a predictive target in selfsupervised learning frameworks. This investigation involved the meticulous calibration of the hyperparameter , integral to the Mixup Feature operation. Such adjustments allowed for the generation of amalgamated feature maps that encompass Sobel edge detection maps, Histogram of Oriented Gradients (HOG) maps, and Local Binary Pattern (LBP) maps, providing a rich, multifaceted representation of visual data. For the empirical application of this novel method, the visual transformer was selected as the principal architecture, due to its proficiency in handling complex visual inputs and its emphasis on critical image regions. This choice was further reinforced by the insights derived from the Masked AutoEncoder (MAE) approach, which illuminated the potential of utilizing partially visible inputs to reconstruct full images, thus enhancing the model's predictive capabilities in a self-supervised context. A denoising self-distillation Masked Autoencoder model for self-supervised learning was developed. This model synthesizes elements from Siamese Networks and Masked Autoencoders, incorporating a tripartite architecture that includes a student network in the form of a masked autoencoder, an intermediary regressor, and a teacher network. The underlying proxy task for this model is the restoration of input images that have been artificially corrupted with random Gaussian noise patches. This is a strategic choice designed to encourage the model to learn robust feature representations by distilling clean signals from noisy inputs. In doing so, the model is trained to reconstruction of the degraded image, effectively teaching it focus on the essence of the visual content. To ensure comprehensive learning, the model harnesses a dual loss function mechanism. One function is calibrated to reinforce the global contextual understanding of the image, thereby enabling the model to grasp the overall structure and scene configuration. Concurrently, the second function is tailored to refine the perception of intricate local details, ensuring that fine visual nuances are not lost in the process of denoising and reconstruction. Through this innovative approach, the model aspires to achieve a delicate balance between the macroscopic comprehension of visual scenes and the meticulous reconstruction of localized details, a balance that is pivotal for sophisticated image analysis tasks in self-supervised learning frameworks. An exhaustive analysis was executed to assess the experimental performance of two innovative self-supervised learning algorithms, specifically applied to three benchmark datasets: Cifar-10, Cifar-100, and STL-10. This study aimed to benchmark these algorithms against existing advanced self-supervised techniques grounded in Masked Image Modeling. In comparison to other state-of-the-art selfsupervised methods based on Masked Image Modeling, the mixed HOG-Sobel feature maps obtained using Mixup showed outstanding performance on Cifar-10 and STL-10 after full fine-tuning, with an average performance improvement of 0.4%. Additionally, the pre-trained model of the Deep Masked Autoencoder (DMAE) was subjected to a rigorous evaluation. When full fine-tuned on the STL-10 dataset, this model demonstrated a modest yet significant edge over the conventional Masked Autoencoder (MAE), exceeding its performance by a margin of 0.1%. This finding shed light on the potential of DMAE in enhancing model accuracy. Moreover, the study revealed that in comparison to traditional self-supervised learning strategies reliant on contrastive learning, the Mixup Feature method emerged as more efficient. It offered the advantage of shortened training durations and negated the requirement for conventional data augmentation methods, thus streamlining the learning process. In conclusion, the two self-supervised learning algorithms introduced in this research contribute to the expanding repertoire of methods for masked image modeling. Their demonstrated effectiveness on benchmark datasets illuminates their potential for broader applications, particularly in larger and more complex datasets. The application of these self-supervised learning algorithms was effectively expanded to encompass the domain of medical image analysis. This extension involved the utilization of self-supervised pre-training on specifically curated medical image datasets. Following this pre-training phase, the model thus developed was then employed for the downstream tasks. Empirical results from this study illustrate that the approach of self-supervised pre-training surpasses the efficacy of direct training methodologies. A notable enhancement in accuracy, exceeding 5%, was observed upon the Full fine-tuning of the model on the two downstream datasets. Data imbalance poses a substantial challenge in medical image analysis, as inadequate representation of specific conditions or features can negatively impact the efficacy of model training and feature extraction. Considering this, the study developed an imbalanced dataset and delved into the robustness of self-supervised pre-trained models in the context of data imbalance. The experimental findings underscore the superior robustness of self-supervised pre-training methods over from scrath trained models in addressing data imbalance issues. Particularly notable is their performance in scenarios with a positive to negative sample ratio of 1:8, where they exhibit enhanced robustness compared to traditional supervised Convolutional Neural Network (CNN) pre-trained models. These results affirm the effectiveness of our proposed self-supervised pre-trained models in tackling dataset imbalance challenges. The notable improvement in the robustness of self-supervised learning algorithms augments their potential as powerful tools in medical image analysis, suggesting a prospective enhancement in accuracy within intelligent assisted diagnostic systems.Документ Відкритий доступ Resistance factors of bacterial nosocomial infections causative agents as background for the modern antimicrobials development(Igor Sikorsky Kyiv Polytechnic Institute, 2023) Wu Lin; Todosiichuk, Tetіana SerhiyivnaWu Lin. Resistance factors of bacterial nosocomial infections causative agents as background for the modern antimicrobials development – qualification scientific work on manuscript rights. Thesis for the degree of Doctor of Philosophy in specialty 091 – Biology. National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute", Kyiv, 2023. The relevance of the research. The era of antibiotics, which began with the discovery of penicillin in the ХХ century, may soon end and humanity will face a challenge to overcome which will have to find new solutions. This challenge is now the "era of antibiotic resistance", caused not only by evolutionary mechanisms of protection of pathogens, but also by many factors of human activity. Particular importance are methods of combating infectious agents when they are in treatment centers and a large number of people can both become infected and be a source of their spread. Such infections are nosocomial (hospital-acquired infection) and are defined by World Organization of Health (WHO) as infections that can infect the patient during treatment in hospital or other health care facilities. The sources of infection in hospitals are not only other patients and staff, but also surfaces, instruments, medical manipulations and operations, which is cause to problems in ensuring proper conditions. However, one of the important factors in the treatment of nosocomial infections is their resistance to many antibiotics used in hospitals at the same time, and as a result “superbug” arise, for which there is no effective counteraction. WHO defines a list of such relevant "superbugs", and almost half of them are included in the already established acronym ESCAPE – Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. It was found that the natural and induced variability of microorganisms that cause inflammatory processes can lead to increased resistance to the action of previously effective antiseptics due to the acquisition of resistance. In addition, the selection of resistant forms of microbial pathogens, can cause low efficiency of therapy, severe disease, long-term treatment or, in some cases, the inability to overcome the infection at all. The mentioned and many other sources, citing WHO and the Center for Disease Control (CDC, USA), state that the need to develop effective antimicrobials against these and other nosocomial infections is a "need of the hour". Obviously, the solution of the problem of overcoming nosocomial infections has many dimensions, including organizational, educational, medical and so on. But these long-term strategies do not remove the urgent task of finding effective antimicrobials or new combinations to treat these severe, often combined infections, right today. This work currently involves a large number of scientists and practitioners, with different approaches to the solvation of the problem. One of the most effective method is undoubtedly the identification of the most vulnerable sites of infectious agents and their application as targets for new drugs. Such vulnerable points of microbial pathogens mainly determine their resistance and pathogenicity, and therefore these factors should be considered more closely to find the target. Therefore, the systematization of scientific data and results in this regard, the analysis and evaluation of the main factors of pathogenicity and resistance of nosocomial and other infections are relevant to determine modern approaches to the development of the latest antimicrobial agents. The goal of the work was to justify systemic approaches to the development of modern antimicrobials based on the analysis of resistance factors of bacterial pathogens of nosocomial infections. To achieve the goal, it was necessary to solve the following problems: - on the basis of theoretical analysis and research, determine the factors and mechanisms of resistance of microbial pathogens, which can be chosen as targets for the action of antiseptics; - to establish the molecular-genetic characteristics of individual representatives of bacterial pathogens that can be used to counter their distribution and to select strategies for antimicrobial therapy; - to propose a method of screening antimicrobial substances for selected pathogenesis targets of pathogenic bacteria, which can be applied in research practice; - to show the prospects of creating effective antiseptics based on microbial products - antibiotics and enzymes of the streptomycete Streptomyces albus; - to develop recommendations for the implementation of methods of monitoring the resistance of nosocomial infections and the principles of modern antimicrobials development. Scientific novelty of the obtained results. The following scientific results were obtained for the first time in the dissertation: - on the basis of genome sequencing and gene annotation of the strain Bacteroides thetaiotaomicron DSMZ 2079 isolated from the patient's blood, the presence of a one- and two-component system for recognizing environmental signals and responding to them was shown, the presence of four homologs of the selftransmitted conjugative transposon CTnDOT, which provides the extension of resistance to tetracycline and erythromycin. The rpoB and tuf genes (in the JHR92_RS03155 and JHR92_RS03195 loci, respectively), which determine the strain's resistance to antibiotics, were identified; - based on the results of the analysis of the 16S rRNA sequences of the isolated clinical strain Pseudomonas oryzihabitans JN 873340 and its comparison with 29 other strains of the species from GenBank, a phylogenetic tree was built and the possibility of identifying their geographical origin by two hypervariable regions V4 and V5 was shown; - the minimum inhibitory concentrations of the new streptofungin antibiotic against C. albicans ATCC 10231 (10 μg/ml), B. subtilis ATCC 6633 (200 μg/ml) and P. aeruginosa ATCC 9027 (500 μg/ml) were established, as well as the absence of toxicity in a wide range of concentrations (from 2.5 to 500 μg/ml), which determines its potential as an active pharmaceutical ingredient; Practical significance of the obtained results. The practical significance of the dissertation consists in solving the scientific and practical problem of developing antimicrobial agents that do not cause the resistance development of microbial pathogens. The developed method for screening inhibitors of sortase A, which is distinguished by the use of cheaper substrates and, accordingly, the possibility of wide application, has been implemented in the practice of the Environmental Comprehensive laboratory of the School of Tropical Medicine of Hainan Medical University, China. (Act of implementation dated September 4, 2023). Developed "Recommendations for the implementation of methods of resistance control of nosocomial infections and principles of development of modern antimicrobial drugs", approved (08.25.23) at the Shupyk National Healthcare University of Ukraine, for use in practice. Proposed as a result of the dissertation research the principles and approaches for the creation of preparations based on biologically active substances with different mechanisms of action were used in the production program of LLC "PHARMA INTERNATIONAL GROUP 2" (Kyiv) in the development of functional cosmetics (Instructions for use dated 20.07.2023 r.). The results of the work were implemented in the teaching of the courses "General microbiology and virology" and "Fundamentals of pharmaceutical production" for students of the specialty 162 "Biotechnology and bioengineering" at the department of industrial biotechnology and biopharmacy of KPI named after Igor Sikorskyi (Act of implementation dated 20.09.2023), as well as for students of the specialty "Tropical medicine" within the framework of lectures and laboratory (practical) classes in the disciplines "Medical immunology", "Pathobiology", "Medical microbiology", "Environmental microbiology ", "Hygienic microbiology", "Environment and health" at Hainan Medical University (China) (Act of implementation dated 01.09.2023). The main provisions of the work are presented in 10 scientific papers: 4 scientific articles included in the Scopus scientometric database, including 2 of them in publications assigned to the 2nd quartile (Q2) in accordance with the SCImago Journal and Country Rank classification (equal to two publications that are counted in accordance with the first paragraph of paragraph 11 of the Resolution of the Cabinet of Ministers of Ukraine dated March 6, 2019 No. 167); 4 articles in specialized foreign periodicals; 2 abstracts at all-Ukrainian and international conferences.Документ Відкритий доступ Simulation of defectoscopic X-ray television systems for non-destructive testing of semiconductor materials(National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, 2007) Slobodian, Nina; Denbnovetsky, StanislavThe work is devoted to increasing effectiveness of X-ray television systems for non-destructive testing on base of selecting by simulation the most profitable functioning regimes of X-ray-electrical signal converter unit for testing of the semiconductor materials. The method of selecting the most profitable X-ray apparatus regime is proposed. The simulation of radiation generation from pulse X-ray tubes is performed. For small-signal approach the end-to-end model of X-ray-electrical signal unit with X-ray vidicon by means of linear digital non-recursive filter is made. The digital non-linear model of such unit is implemented. The end-to-end model of converter with CCD-matrix is developed and applied. Good correspondence between the results obtained in the framework of proposed model and by performed experiments is achieved.Документ Відкритий доступ Автоматизация процесса управления городским хозяйством(2017) Губский, Андрей Николаевич; Стенин, Александр Африканович; Технической кибернетики; Информатики и вычислительной техники; Национальной технический университет Украины «Киевский политехнический институт имени Игоря Сикорского»Документ Відкритий доступ Автоматизація процесу керування багатокамерними печами випалювання вуглеграфітових виробів(2020) Коротинський, Антон Петрович; Жученко, Олексій АнатолійовичДокумент Відкритий доступ Автоматизація процесу керування усталеним рухом антропоморфного крокуючого апарата(2016) Гуменний, Дмитро Олександрович; Ткач, Михайло Мартинович; Кафедра технічної кібернетики; Факультет інформатики та обчислювальної техніки; Національний технічний університет України "Київський політехнічний інститут"Документ Відкритий доступ Автоматизація процесу керування усталеним рухом антропоморфного крокуючого апарата(НТУУ «КПІ», 2016) Гуменний, Дмитро Олександрович; Кафедра технічної кібернетики; Факультет інформатики та обчислювальної техніки; Національний технічний університет України «Київський політехнічний інститут»Документ Відкритий доступ Автоматизація процесу керування формуванням вуглецевих виробів(2020) Хібеба, Микола Григорович; Жученко, Анатолій ІвановичДокумент Відкритий доступ Автоматизація процесу стабілізації програмного руху безпілотного літального апарату (БПЛА)(2019) Солдатова, Марія Олександрівна; Ткач, Михайло МартиновичДокумент Відкритий доступ Автоматизація процесу стабілізації програмного руху безпілотного літального апарату (БПЛА)(КПІ ім. Ігоря Сікорського, 2019) Солдатова, Марія ОлександрівнаДокумент Відкритий доступ Автоматизація процесу управління міським господарством(КПІ ім. Ігоря Сікорського, 2017) Губський, Андрій Миколайович; Технічної кібернетики; Інформатики та обчислювальної техніки; Національний технічний університет України «Київський політехнічний інститут імені Ігоря Сікорського»Документ Відкритий доступ Автоматизація процесів керування прогріванням паперового полотна у сушильній частині папероробної машини(КПІ ім. Ігоря Сікорського, 2017) Черьопкін, Євгеній СергійовичДокумент Відкритий доступ Автоматизація процесів керування прогріванням паперового полотна у сушильній частині папероробної машини(2017) Черьопкін, Євгеній Сергійович; Жученко, Анатолій Іванович