|
|
Signature verification using neural network-based algorithms
Čírtek, Petr ; Kiac, Martin (referee) ; Myška, Vojtěch (advisor)
Signature is one of the most used biometrics in banking and contracting therefore is important to verificate signature authenticity. Verification can be done with the help of a forensic specialist or, thanks to the rise of advanced technology, with the help of a computing technology. The purpose of this thesis is to develop methods for signature verification using neural networks for Czech type of signature and to find out if adding manual extracted features to convolutional analysis could improve these methods. Neural networks seek to replicate the functioning of human brain, consisting of input neurons, several hidden layers and output neurons. Neural networks are one of the most popular artificial intelligence technologies for image analysis and classification. The proposed methods in this thesis work on the principles of convolutional networks. The first proposed method consist of three convolutional layers which extract important features from image of signature and pass them to fully connected classifier layer. This determines whether the signature is genuine or forgery. Also for this method there were created two functions which can interpret it's decision-making. The second method, siamese neural network, unlike the first, does not work with signatures independently, but uses a reference signature image to determine authenticity. The basis of this method is to extract features with convolutional analysis from both the reference signature and the signature to be authenticated. These features are then concatenated and passed to the clasificator. A Czech dataset was created to train models that would verify the Czech type of signatures. From the experiments, it was found that the addition of manualy extracted features has the potential to improve the prediction accuracy of methods based on convolutional image analysis. 3 models were trained, which can verify the Czech type of signatures with an accuracy higher than 80 \%, namely: the model of the convolutional neural network method with discrete wavelet transformation feature, which was trained on the Czech dataset, the model of the same method trained on the CEDAR dataset with number of strokes as added feature and a siamese convolutional neural network method model trained on the Czech dataset of signatures with the tri-surface feature.
|
|
|
Reconstruction of corrupted audio signals using deep unrolling
Kment, František ; Myška, Vojtěch (referee) ; Mokrý, Ondřej (advisor)
The thesis deals with the problem of audio signal restoration using traditional optimization methods combined with deep unrolling methods. An optimization task for filling in missing sections of the audio signal was formulated, and the proximal algorithm FISTA was chosen and subsequently implemented. Furthermore, three unrolled variants of the algorithm (Unrolled Fista Net) were implemented, two of which were further optimized using tests on a selected test dataset and trained on the Nsynth dataset. The results of the trained networks were compared both with competing methods and the original untrained variant of the algorithm. The comparison was made using objective metrics (MSE, SNR, PEAQ, PEMO-Q) and a subjective listening test.
|
|
|
The use of federated learning in the field of security on Android OS
Szüč, Martin ; Myška, Vojtěch (referee) ; Michálek, Jakub (advisor)
This thesis explores the use of federated learning in the context of cybersecurity, specifically in detecting phishing attacks via email on the Android operating system. The~theoretical part of the thesis deals with concepts of federated learning, machine learning, and various phishing techniques. The main goal of the practical part is to design and implement a mobile application that uses federated learning to train machine learning models. This application is designed to detect phishing emails while ensuring that the content of the emails is not sent to a central server, thereby protecting users' sensitive data. The~thesis includes the design of the application architecture, integration of Python modules, processing and classification of emails, and implementation of federated learning. The results of the application testing demonstrate the effectiveness of the proposed solution in detecting phishing attacks while also highlighting the privacy benefits provided by federated learning.
|
|
|
Optimization of control using reinforcement learning on the Robocode platform
Pastušek, Václav ; Myška, Vojtěch (referee) ; Burget, Radim (advisor)
This master's thesis focuses on optimizing the control of a tank robot in the Robocode environment using reinforcement learning. The complexity of this problem falls into the EXPSPACE class, presenting a challenge that cannot be underestimated. The theoretical part of the thesis meticulously examines the Robocode platform, concepts of reinforcement learning, and relevant algorithms, while the practical part focuses on optimizing the agent, implementing reinforcement learning algorithms, and creating a user-friendly interface for easy training and testing of models. A total of 64 models were trained and tested as part of the thesis, with their data and parameters compared and presented in accompanying databases and graphs. The best results in terms of average hits per episode were achieved by models labeled v0.8.0 and v1.0.0. The first model exhibited a certain ability to evade shots, while the second model showed more successful hits.
|
|
|
COVID-19 disease classification based on analysis of chest X-rays
Šteflík, Dominik ; Kiac, Martin (referee) ; Myška, Vojtěch (advisor)
This diploma thesis addresses the development and evaluation of artificial intelligence algorithms for classifying COVID-19 disease from chest X-ray images. Given the severity and impact of the COVID-19 pandemic on the global population, the ability to rapidly and accurately diagnose diseases from radiographic images has become critical. This study synthesizes current advancements in image processing and deep learning to evaluate the application of several novel classification methods in practice. Using a dataset obtained from a Czech medical environment, these methods are analyzed and validated in order to examine their effectiveness and accuracy in real life scenarios. The methods chosen for this study, COVID-Net, DarkCovidNet, and CoroNet, were selected due to their availability, widespread use and proven effectiveness in the field. The core of the thesis is the design of a convolutional neural network tailored to extract and learn from the subtle features present in X-ray images indicative of COVID-19. This initiative confronted significant challenges posed by variable acquisition parameters of X-ray images, which can substantially affect diagnostic accuracy. The uniformity of these parameters is crucial for reliable analysis, underscoring the importance of rigorous preprocessing techniques. In response, advanced normalization, contrast adjustment, and augmentation procedures were implemented to standardize the input data. The convolutional network itself employs a series of convolutional, pooling, and fully connected layers, optimized to handle the nuanced variations present in medical imaging data. Notably, the network architecture incorporates an attention mechanism, implemented through a Squeeze-and-Excitation block, to dynamically adjust the importance of different channels in the input image. By integrating these elements, the network model is trained to focus on significant features within the X-ray images, allowing it to distinguish subtle indicators of COVID-19 effectively. Furthermore, this work discusses the potential of integrating these AI-driven diagnostic tools into existing healthcare infrastructures to enhance early detection and treatment of COVID-19. The findings indicate that leveraging artificial intelligence in medical imaging can substantially aid in managing and controlling disease outbreaks, ultimately contributing to better health outcomes.
|
|
|
Recurrent Neural Network for Text Classification
Myška, Vojtěch ; Kolařík, Martin (referee) ; Povoda, Lukáš (advisor)
Thesis deals with the proposal of the neural networks for classification of positive and negative texts. Development took place in the Python programming language. Design of deep neural network models was performed using the Keras high-level API and the TensorFlow numerical computation library. The computations were performed using GPU with support of the CUDA architecture. The final outcome of the thesis is linguistically independent neural network model for classifying texts at character level reaching up to 93,64% accuracy. Training and testing data were provided by multilingual and Yelp databases. The simulations were performed on 1200000 English, 12000 Czech, German and Spanish texts.
|
|
|
Client-server search engine for optimal public transport routes
Brát, Daniel ; Hošek, Jiří (referee) ; Myška, Vojtěch (advisor)
This bachelor thesis deals with the design and implementation of a client-server based program for finding the optimal path in public transport networks. Theoretical basics describe the Traveling salesman problem and mention some methods for finding its solution. This chapter is followed by a theoretical design of the application itself. Next chapter describes chosen technologies to be used. The last chapter describes the realization of the application itself. Reached goals are described in the conclusion.
|
|
|
Segmentation of multiple sclerosis lesions using deep neural networks
Sasko, Dominik ; Myška, Vojtěch (referee) ; Kolařík, Martin (advisor)
Hlavným zámerom tejto diplomovej práce bola automatická segmentácia lézií sklerózy multiplex na snímkoch MRI. V rámci práce boli otestované najnovšie metódy segmentácie s využitím hlbokých neurónových sietí a porovnané prístupy inicializácie váh sietí pomocou preneseného učenia (transfer learning) a samoriadeného učenia (self-supervised learning). Samotný problém automatickej segmentácie lézií sklerózy multiplex je veľmi náročný, a to primárne kvôli vysokej nevyváženosti datasetu (skeny mozgov zvyčajne obsahujú len malé množstvo poškodeného tkaniva). Ďalšou výzvou je manuálna anotácia týchto lézií, nakoľko dvaja rozdielni doktori môžu označiť iné časti mozgu ako poškodené a hodnota Dice Coefficient týchto anotácií je približne 0,86. Možnosť zjednodušenia procesu anotovania lézií automatizáciou by mohlo zlepšiť výpočet množstva lézií, čo by mohlo viesť k zlepšeniu diagnostiky individuálnych pacientov. Našim cieľom bolo navrhnutie dvoch techník využívajúcich transfer learning na predtrénovanie váh, ktoré by neskôr mohli zlepšiť výsledky terajších segmentačných modelov. Teoretická časť opisuje rozdelenie umelej inteligencie, strojového učenia a hlbokých neurónových sietí a ich využitie pri segmentácii obrazu. Následne je popísaná skleróza multiplex, jej typy, symptómy, diagnostika a liečba. Praktická časť začína predspracovaním dát. Najprv boli skeny mozgu upravené na rovnaké rozlíšenie s rovnakou veľkosťou voxelu. Dôvodom tejto úpravy bolo využitie troch odlišných datasetov, v ktorých boli skeny vytvárané rozličnými prístrojmi od rôznych výrobcov. Jeden dataset taktiež obsahoval lebku, a tak bolo nutné jej odstránenie pomocou nástroju FSL pre ponechanie samotného mozgu pacienta. Využívali sme 3D skeny (FLAIR, T1 a T2 modality), ktoré boli postupne rozdelené na individuálne 2D rezy a použité na vstup neurónovej siete s enkodér-dekodér architektúrou. Dataset na trénovanie obsahoval 6720 rezov s rozlíšením 192 x 192 pixelov (po odstránení rezov, ktorých maska neobsahovala žiadnu hodnotu). Využitá loss funkcia bola Combo loss (kombinácia Dice Loss s upravenou Cross-Entropy). Prvá metóda sa zameriavala na využitie predtrénovaných váh z ImageNet datasetu na enkodér U-Net architektúry so zamknutými váhami enkodéra, resp. bez zamknutia a následného porovnania s náhodnou inicializáciou váh. V tomto prípade sme použili len FLAIR modalitu. Transfer learning dokázalo zvýšiť sledovanú metriku z hodnoty približne 0,4 na 0,6. Rozdiel medzi zamknutými a nezamknutými váhami enkodéru sa pohyboval okolo 0,02. Druhá navrhnutá technika používala self-supervised kontext enkodér s Generative Adversarial Networks (GAN) na predtrénovanie váh. Táto sieť využívala všetky tri spomenuté modality aj s prázdnymi rezmi masiek (spolu 23040 obrázkov). Úlohou GAN siete bolo dotvoriť sken mozgu, ktorý bol prekrytý čiernou maskou v tvare šachovnice. Takto naučené váhy boli následne načítané do enkodéru na aplikáciu na náš segmentačný problém. Tento experiment nevykazoval lepšie výsledky, s hodnotou DSC 0,29 a 0,09 (nezamknuté a zamknuté váhy enkodéru). Prudké zníženie metriky mohlo byť spôsobené použitím predtrénovaných váh na vzdialených problémoch (segmentácia a self-supervised kontext enkodér), ako aj zložitosť úlohy kvôli nevyváženému datasetu.
|
|
|
Image data segmentation using deep neural networks
Hrdý, Martin ; Myška, Vojtěch (referee) ; Kiac, Martin (advisor)
The main aim of this master’s thesis is to get acquainted with the theory of the current segmentation methods, that use deep learning. Segmentation neural network that will be capable of segmenting individual instances of the objects will be proposed and created based on theoretical knowledge. The main focus of the segmentation neural network will be segmentation of electronic components from printed circuit boards.
|
|
|
Object detection in video using neural networks
Mikulský, Petr ; Sikora, Pavel (referee) ; Myška, Vojtěch (advisor)
This diploma thesis deals with the detection of moving objects in a video recording using neural networks. The aim of the thesis was to detect road users in video recordings. Pre-trained YOLOv5 object detection model was used for a practical part of the thesis. As part of the solution, an own dataset of traffic road video recordings was created and annotated with following classes: a car, a bus, a van, a motorcycle, a truck and a trailer truck. Final version of this dataset comprise 5404 frames and 6467 annotated objects in total. After training, the YOLOv5 model achieved 0.995 mAP, 0.995 precision and 0.986 recall on the dataset. All steps leading to the final form of the dataset are described in the conclusion chapter.
|
guest ::
RSS feed.