This thesis deals with image classification based on genetic programming and coevolution. Genetic programming algorithms make generating executable structures possible, which allows us to design solutions in form of programs. Using coevolution with the fitness prediction lowers the amount of time consumed by fitness evaluation and, therefore, also the execution time. The thesis describes a theoretical background of evolutionary algorithms and, in particular, cartesian genetic programming. We also describe coevolutionary algorithms properties and especially the proposed method for the image classifier evolution using coevolution of fitness predictors, where the objective is to find a good compromise between the classification accuracy, design time and classifier complexity. A part of the thesis is implementation of the proposed method, conducting the experiments and comparison of obtained results with other methods.

Ne vždy je jednoduché získání dostatečně velké a kvalitní datové sady s obrázky zřetelných artefaktů, ať už kvůli nedostatku ze strany zdroje dat nebo složitosti tvorby anotací. To platí například pro radiologii, nebo také strojírenství. Abychom mohli využít moderní uznávané metody strojového učení které se využívají pro klasifikaci, segmentaci a detekci defektů, je potřeba aby byla datová sada dostatečně velká a vyvážená. Pro malé datové sady čelíme problémům jako je přeučení a slabost dat, které způsobují nesprávnou klasifikaci na úkor málo reprezentovaných tříd. Tato práce se zabývá prozkoumáváním využití generativních sítí pro rozšíření a vyvážení datové sady o nové vygenerované obrázky. Za použití sítí typu Conditional Generative Adversarial Networks (CGAN) a heuristického generátoru anotací jsme schopni generovat velké množství nových snímků součástek s defekty. Pro experimenty s generováním byla použita datová sada závitů. Dále byly použity dvě další datové sady keramiky a snímků z MRI (BraTS). Nad těmito dvěma datovými sadami je provedeno zhodnocení vlivu generovaných dat na učení a zhodnocení přínosu pro zlepšení klasifikace a segmentace.

The aim of this work was to solve problems with excessive humidity in buildings using data analysis. The theoretical part of the work deals with impacts of excessive humidity on the health of building occupants and also the condition of the building structure. Data mining methods including classification, prediction, and clustering are described together with model evaluation and selection. The practical part focuses on hardware platform description and measurement scenarios. Key parameters affecting indoor relative humidity are indoor and outdoor temperature and outdoor relative humidity. The long-term measurement of the mentioned parameters was performed using the set of sensors and BeeeOn system. Measured data was used to design a system for event detection related to a humidity change. The approach to air change regulation in the room was based on natural ventilation.

The aim of this work is to describe the past, present and expected future of the airport in Brno. An analysis was carried out that was sent as a reminder of the airport's potential, which should be revealed in the near future. The appendix describes airport development methods with similar characteristics used by other managers

The master's thesis deals with the design and time optimization of the method designed to determine the classification of LED lamps due to their resistance to power supply fluctuations. Classification of LED lamps in the classification scale and their labelling by the classification index is the task of informing in a simple way those interested in the ability of a concrete light source to function properly, without disturbing flickering, in conditions of electromagnetic interference.

This master's thesis focuses on detection of pathological vertebrae in spinal CT utilized by machine learning. Theoretical part describes anatomy of the spine and occurrence of pathologies in CT image data, contains an overview of existing methods intended for automated detection of pathological vertebrae. Practical part devotes to design a computer aided detection systems to identify pathological vertebrae and to classify a type of pathology. Designed classification system is based on using neural network, which performs classification step and on principal component analysis (PCA), which is used to reducing the original number of observation features. For completing this task were used real data. Conclusion contains evaluation of obtained results.

Tato práce se zabývá problematikou automatizované klasifikace a rozpoznávání bakterií po získání jejich DNA procesem sekvenování. V rámci této práce je navržena a popsána nová metoda klasifikace založená na základě segmentu 16S rRNA. Představený princip je vytvořen podle stromové struktury taxonomických kategorií a používá známé algoritmy strojového učení pro klasifikaci bakterií do jedné ze tříd na nižší taxonomické úrovni. Součástí práce je dále implementace popsaného algoritmu a vyhodnocení jeho přesnosti predikce. Přesnost klasifikace různých typů klasifikátorů a jejich nastavení je prozkoumána a je určeno nastavení, které dosahuje nejlepších výsledků. Přesnost implementovaného algoritmu je také porovnána s několika existujícími metodami. Během validace dosáhla implementovaná aplikace KTC více než 45% přesnosti při predikci rodu na datových sadách BLAST 16S i BLAST V4. Na závěr je zmíněno i několik možností vylepšení a rozšíření stávající implementace algoritmu.

This thesis deals with a measurement and interpretation of NIR spectra of bacterial cells and design of biosensor using this analytical technique. In the first chapter, there is introduction of current state of knowledge in the field of NIR spectroscopy in microbiology and technology of fiber optic biosensors. The summary of this chapter shows that NIR is a suitable technique for direct molecular analysis of bacteria, but it suffers from low sensitivity and insufficient interpretation of bacterial spectra. In the next part of the thesis, there is a theoretical background of spectral analysis techniques and technology of fiber optic sensors. In the practical part of this work, there is suggested the elimination of disadvantages of NIR spectroscopy in microbiology by a series of experiments used for interpretation of NIR spectra of bacteria and design of fiber optic sensor to increase sensitivity of this technique. In this work, spectral regions important for the identification of bacterial strains were determined and partially interpreted and the sensor for bacterial analysis capable of classifying strains based on 105 captured cells was designed. Therefore, the objectives of this work were fulfilled.

This thesis attempts to find an alternative method for biometric identification using retinal images. First part is focused on the introduction to biometrics, human eye anatomy and methods used for retinal biometry. The essence of neural networks and deep learning methods is described as it will be used practically. In the last part of the thesis a chosen identification algorithm and its implementation is described and the results are presented.

This diploma thesis focuses on the automated diagnosis of hypokinetic dysarthria in the multilingual speech corpus, which is a motor speech disorder that occurs in patients with neurodegenerative diseases such as Parkinson’s disease. The automatic speech recognition approach to diagnosis is based on the acoustic analysis of speech and subsequent use of mathematical models. The popularity of this method is on the rise due to its objectivity and the possibility of working simultaneously on different languages. The aim of this work is to find out which acoustic parameters have high discriminative power and are universal for multiple languages. To achieve this, a statistical analysis of parameterized speech tasks and subsequent modelling by machine learning methods was used. The analyses were performed for Czech, American English, Hungarian and all languages together. It was found that only some parameters enable the diagnosis of the hypokinetic disorder and are, at the same time, universal for multiple languages. The relF2SD parameter shows the best results, followed by the NST parameter. When classifying speakers of all the languages together, the model achieves accuracy of 59 % and sensitivity of 72 %.