Title: Converting Prose into Poetry with Neural Networks Author: Memduh Gokirmak Institute: Institute of Formal and Applied Linguistics Supervisor: Martin Popel, Institute of Formal and Applied Linguistics Abstract: We present here our attempts to create a system that generates poetry based on a sequence of text provided to it by a user. We explore the use of machine translation and language model technologies based on the neural network architecture. We use different types of data across three languages in our research, and employ and develop metrics to track the quality of the output of the systems we develop. We find that combining machine translation techniques to generate training data to this end with fine-tuning of pre-trained language models provides the most satisfactory generated poetry. Keywords: poetry machine translation language models iii

The aim of this work is to design methods of fingerprint morphing, subsequently to implement the proposed methods and to test their effectiveness by breaking into real verification systems made by Innovatrics and Neurotechnology, two prominent companies in biometrics. Methods for morphing were designed mainly based on the article On the Feasibility of Creating Double-Identity Fingerprints where the part of the preparation of fingerprints on the input was modified and an adjustment of the output was made so that it is able to work with fingerprint images taken under different conditions. Fingerprints created by morphing were recognized as genuine in 41.72 % of cases when compared to the default threshold. The main benefit of this work is the created application for generating fingerprints using morphing and to point out that this method is a\,threat to existing systems used for fingerprint recognition.

The aim of this master's thesis has been to develop and optimize the method of UV- photochemical vapor generation of cadmium volatile compounds for atomic fluorescence spectrometry. Two configurations with different materials wrapped around the low-pressure mercury vapor lamp were tested. I experimentally determined optimal conditions for both systems; the optimalized parametres included selection of photochemical reagent and its concentration, flow rate of liquids and gases, and the lenghth of the reaction coil. After finding the optimal parametres I determined the figures of merit of the method. I found from the measured data that UV-photoreactor with the quartz capillary provided lower limits of detection and thus was more suitable for generation of cadmium volatile compounds. With this arrangement and using the ferrous sulphate heptahydrate as the chemical modifier I obtained limit of detection 2,0 µg·l-1 , limit of determination 7,0 µg·l-1 , linear dynamic rate LOD-50 µg·l-1 and repeatibility 0,35 %. I also carried out the interference study in my thesis and determined the generation effeciency of cadmium volatile compounds. The interference study shows the influence of mineral acids (HCl, H2SO4, HNO3), their salts (NaNO2, NaNO3) and transition metals (Co, Ni, Cu). Based on literature review I also...

Games get boring when they start repeating themselves and do not offer players new content. Procedural content generation (PCG) is increasingly used to generate this content. PCG-based game design decreases the need to have a human designer or a writer to generate the content. Algorithmic creation of game content can augment the creativity of human designers and this makes it possible small so-called indie teams to create the content for their game without the big resources. In this work, the field of PCG is introduced. Application of PCG is shown through a mobile game implementation. The implementation details of the mobile game Rush Hour will be presented that makes use of Foursquare, Twitter and Mapbox APIs, which eases the content creation using open data as the input of PCG.

Title: Novel Methods for Natural Language Generation in Spoken Dialogue Systems Author: Ondřej Dušek Department: Institute of Formal and Applied Linguistics Supervisor: Ing. Mgr. Filip Jurčíček, Ph.D., Institute of Formal and Applied Linguistics Abstract: This thesis explores novel approaches to natural language generation (NLG) in spoken dialogue systems (i.e., generating system responses to be presented the user), aiming at simplifying adaptivity of NLG in three respects: domain portability, language portability, and user-adaptive outputs. Our generators improve over state-of-the-art in all of them: First, our gen- erators, which are based on statistical methods (A* search with perceptron ranking and sequence-to-sequence recurrent neural network architectures), can be trained on data without fine-grained semantic alignments, thus simplifying the process of retraining the generator for a new domain in comparison to previous approaches. Second, we enhance the neural-network-based gener- ator so that it takes preceding dialogue context into account (i.e., user's way of speaking), thus producing user-adaptive outputs. Third, we evaluate sev- eral extensions to the neural-network-based generator designed for producing output in morphologically rich languages, showing improvements in Czech generation. In...

The thesis deals with a computer modelling by means of the commercial software COMSOL Multiphysics. The thesis is structured into three topical sections. The first part is dedicated to the programme characteristics and its development. In the second part the brief characteristics of working with the programme COMSOL Multiphysics is presented which should make the user´s first steps in working with software easier. The last part of thesis shows the specific demonstration of the created multiple physics task taken from the field of plasma physics. This task is solved by means of continuous simulation of a computer modelling based on the drift-diffusion approximation of low temperature plasma. The proposal of the paper for the proceedings of the conference ,,Technical Computing Prague 2012" is also a part of this thesis.

The goal of this thesis is to design and implement a program capable of automatically generating synthetic images of eye retinas. The generated images should be similar to those of real retinas, which are hard to obtain, so they could be used for development of various algorithms, which work with eye retina images in their place. This thesis describes anatomic properties of the eye retina, ways to take images of it and also usage of eye retina recognition in biometric and medicinal applications. Design of the program and the way in which individual parts of the retina are assembled together into the final image is also explained in this thesis. These individual parts are created using procedural textures in separate layers of the image. Next chapter of this work describes implementation details of the program. The conclusion then experimentally verifies suitability of the generated images for algorithmic processing.

This thesis deals with the design and implementation of Android application development tool based on model driven software development. The first part of the thesis is focused on general software development and next part on software development based on model driven development and executable UML. In next part Android platform, methods of Android application development and existing MDD tools are described. This thesis continues with the design of my own MDD tool for the creation of Android applications. The designed tool is realized like Gradle plugin and independent development environment using thisplugin. The designed tool is based on fUML and ALF language. The features and options of development tool are demonstrated by creation of example application.

The diploma thesis deals with an automatic code generation for PLC from Siemens. The first part focuses on currently available tools for code generation and options of C# library Siemens TIA Openness. Furthermore, a design for the project structure in the TIA portal is created. Finally, a user application for automatic code generation in TIA portal v15 is designed and created, especially suitable for larger projects that could be divided into individual stations.

Pokrok při zdokonalování klasifikačních algoritmů je zpomalován nedostatkem dat potřebných pro testování. Reálná data je obtížné získat z důvodu bezpečnosti a ochrany citlivých informací. Existují však generátory syntetických sad pravidel, jako například ClassBench-ng. K vyhodnocení správného fungování, propustnosti, spotřeby energie a dalších vlastností klasifikačních algoritmů je zapotřebí také vhodný síťový provoz. Tématem této práce je tvorba takového generátoru síťového provozu, který by umožnil testování těchto vlastností v kombinaci s IPv4, IPv6 a OpenFlow1.0 pravidly vygenerovanými ClassBench-ng. Práce se zabývá různými způsoby, jak toho dosáhnout, které vedly k vytvoření několika verzí generátoru. Vlastnosti jednotlivých verzí byly zkoumány řadou experimentů. Implementace byla provedena pomocí jazyku Python. Nejvýznamnějším výsledkem je generátor, který využívá principů několika zkoumaných přístupů k dosažení co nejlepších vlastností. Dalším přínosem je nástroj, který bylo nutné vytvořit pro analýzu užitých sad klasifikačních pravidel a vyhodnocení vlastností vygenerovaného síťového provozu.