|
|
Obstacle Avoidance in UAVs: Using a Bug-Inspired Algorithm and Neural Network-Based RGB Camera Collision Prediction
Raichl, Petr ; Marcoň, Petr ; Janoušek, Jiří
Unmanned Aerial Vehicles (UAVs) are increasingly deployed in complex environments for various applications, necessitating advanced obstacle avoidance capabilities to ensure safety and mission success. Inspired by the simplicity and effectiveness of biological navigation strategies, this study introduces a novel approach to UAV obstacle avoidance, leveraging the principles of the bug algorithm combined with the predictive power of neural networks. We propose a hybrid model that integrates a lightweight neural network to predict potential collisions in real-time. Our methodology employs a two-stage process: first, the neural network assesses the immediate risk of collision; second, the bug algorithm-inspired decision-making process determines the UAV’s maneuvering actions to navigate without crashing to obstacles. The approach was tested both in simulation and real outdoor experiments.
|
|
|
The impact of AI tools on higher education students
Lacina, Ondřej ; Soukup, Petr (advisor) ; Remr, Jiří (referee)
In the world of rapidly evolving technologies, the last two years have seen a massive expansion of artificial intelligence. We are therefore all encountering emerging AI tools in our everyday lives. The thesis titled The Impact of AI Tools on Higher Education Students focuses on the ways in which students of Czech universities approach these new emerging technologies. After explaining the notion of artificial intelligence, the fundamental breakdown of the tools is outlined. The possible good and negative elements of these new instruments are also discussed. For the research, a mixed research design is adopted, which separates the analysis into two sections using the explanatory sequential design's processes. In the first step, a quantitative questionnaire survey was undertaken, with respondents divided proportionally into ten quotas based on the ISCED-F 2013 category. Semi-structured interviews were used in the second stage to augment the questionnaire survey. The analysis resulted in the key trends in the use of AI tools among students of Czech universities. High focus is given on the frequency of use, preferred tools, usual purposes of usage in academic, private, and professional life, faith in the generated information, and worries about potential negative consequences. The usual user groups for...
|
|
|
Modern vibrodiagnostics of machines and evaluation of datasets by neural networks
Koníček, Tomáš ; Holoubek, Tomáš (referee) ; Hammer, Miloš (advisor)
This Master‘s thesis focuses on technical diagnostics with an emphasis on vibrodiagnostics of machines and equipment. The aim is to carry out research on vibration monitoring using modern on-line systems and to investigate the possibilities of processing the acquired data files using neural networks. Vibration monitoring from Siemens SIPLUS CMS is analyzed, including a description of individual hardware and software components. The work also focuses on machine diagnostics using a real model equipped with the SIPLUS CMS system in cooperation with the SIMATIC S7-1200 programmable automaton. The obtained data will be transferred via the FTP protocol for further processing in the Matlab program. Neural network models will be designed and used, which will be trained on the measured data. Convolutional neural network model will be used. The results will be evaluated and a conclusion will be drawn.
|
|
|
Deepfake Detection in Video Samples
Krumpholc, Jan ; Veigend, Petr (referee) ; Lapšanský, Tomáš (advisor)
V posledních letech si můžeme všimnout nárůstu internetových podvodů a podvrhů. Počínaje snadno odhalitelnými případy, jako je phishing a falešné reklamy, přes sociální inženýrství a dezinformační kampaně a konče útoky pomocí umělé inteligence: Syntetická media, a obzvláště deepfakes. Tyto útoky jsou velmi efektivní, protože je obtížné ověřit pravost média pro běžného uživatele, a také díky nárustu dostupnosti a efektivity těchto nástrojů pro veřejnost v posledních letech. Tato bakalářská práce je zaměřena na video deepfakes: Jaké metody se používají k jejich tvorbě, jaké jsou jejich slabé stránky a hlavně, jak tyto slabé stránky najít a rozhodnout, zda je médium deepfake či nikoli. Budeme analyzovat aktuálně nejmodernější metody detekce deepfakes, jaké jsou jejich silné a slabé stránky, a vyvineme možné nové metody detekce. Nakonec porovnáme výsledky s aktuálními řešeními a vyhodnotíme výsledek.
|
|
|
Machine Learning of Representations in Genetic Programming
Pomykal, Šimon ; Piňos, Michal (referee) ; Sekanina, Lukáš (advisor)
The aim of this thesis is to become acquainted with machine learning methods that are used for the automatic design of representations. Specifically, the work focuses on deep learning in the field of genetic programming (GP). Image processing is chosen as a case study, particularly noise reduction methods. By combining the acquired knowledge, a new representation is proposed, intended to replace the syntactic tree in the GP algorithm. This method is obtained using a transformer-type neural network. In conclusion, a modified version of GP that works with the new representation is created. This variant is compared with the original GP using the traditional representation in several experiments.
|
|
| |
|
|
Reprezentace síťových toků s využitím neuronových sítí
Pycz, Lukasz ; Jeřábek, Kamil (referee) ; Poliakov, Daniel (advisor)
This thesis explores the application of self-supervised learning (SSL) methods such as data masking, data order shuffling, and contrastive learning, to extract meaningful representations from network flow data, specifically using the CESNET TLS22 dataset from CESNET DataZoo. The main goal is to develop a robust model that improves the understanding and analysis of network flows through effective representation learning without relying on labeled data. The research utilizes the PyTorch computational framework for designing, training, and evaluating the performance of the model.
|
|
|
Reverse engineering of an audio mix using neural networks
Čermák, Jerguš ; Mokrý, Ondřej (referee) ; Rajmic, Pavel (advisor)
This bachelor's thesis focuses on the use of neural network algorithms to determine the parameters of signal processors used in the mixing of audio tracks. The thesis presents linear signal processors such as \emph{Gain}, \emph{Pan}, \emph{Filter}, and \emph{Reverb}, which are commonly used to process audio signal and to produce a stereo mix of the audio recording. These processors are subsequently used within the neural network model as layers, implemented using the DDSP library, aimed at predicting the parameters used in the mix, given the knowledge of the input tracks and the target mix. Resultantly, stereo mixdowns were created, and their parameters were estimated using two neural network models. The results were evaluated using both objective measurements and subjective methods (listening test).
|
|
|
Use of Diffusion Models in Deepfakes
Trúchly, Dominik ; Malinka, Kamil (referee) ; Lapšanský, Tomáš (advisor)
Deepfake je typ syntetického média vytvoreného pomocou sofistikovaných algoritmov strojového učenia, najmä hlbokých neurónových sietí. Ako príklad možno uviesť generatívne adverzné neurónové siete (GAN), ktoré sú schopné generovať obrázky, ktoré sú pre bežných jednotlivcov takmer nemožné odlíšiť od skutočnej reality. V dôsledku toho boli vyvinuté algoritmy detekcie hlbokých falošných správ, ktoré riešia tento rastúci problém. Tieto algoritmy využívajú pokročilé techniky strojového učenia a analyzujú rôzne funkcie v rámci obrázkov a videí, aby identifikovali nezrovnalosti alebo anomálie svedčiace o manipulácii. Táto práca skúma aplikáciu difúznych modelov, bežne používaných v digitálnom spracovaní obrazu na zvýšenie kvality obrazu znížením šumu a rozmazania, pre posilňovanie realizmu deepfakes. Využitím týchto modelov testujeme ich efekt na odhaľovanie deepfakes obrázkov pomocou deepfake detektorov.
|
|
|
Simulation of Biological Processes Using Asynchronous Cellular Automata and Machine Learning
Kališ, Vojtěch ; Bidlo, Michal (referee) ; Fritz, Karel (advisor)
Tato práce zkoumá spojení asynchronních celulárních automatů a technik strojového učení pro simulaci komplexních biologických procesů. Jejím hlavním zaměřením je předvést vrozený potenciál výpočetního rámce konstruovaného spojením paralelismu aktualizačního modelu asynchronních celulárních automatů s prediktivními schopnostmi algoritmů strojového učení. Tato studie si klade za cíl demonstrovat kvality takového hybridního přístupu implementací tří matematických modelů celulárních automatů s rostoucí složitostí—tj., seřezeny podle stupně složitosti, Conwayova Hra Života, SmoothLife a Lenia—ve své základní formě a následnou integrací strojového učení do funkce dvou posledně jmenovaných, po čemž následuje porovnání výsledků obou přístupů.
|
guest ::
