|
|
Dynamic Programming in Graph Algorithms
Biloš, Martin ; Křivka, Zbyněk (referee) ; Burgetová, Ivana (advisor)
This work is about graph algorithms, their use and the benefit of the optimization method of dynamic programming. This benefit is show to the user via the graphic application. Graph algorithms find use in many sectors of human activity. They are used in packet routing or, for example, navigation. There are three methods of graph algorithms used in this work. This problems are solved with classic and dynamic way and measured data are compared.
|
|
|
Machine Learning Concepts for Categorization of Objects in Images
Hubený, Marek ; Honec, Peter (referee) ; Horák, Karel (advisor)
This work is focused on objects and scenes recognition using machine learning and computer vision tools. Before the solution of this problem has been studied basic phases of the machine learning concept and statistical models with accent on their division into discriminative and generative method. Further, the Bag-of-words method and its modification have been investigated and described. In the practical part of this work, the implementation of the Bag-of-words method with the SVM classifier was created in the Matlab environment and the model was tested on various sets of publicly available images.
|
|
|
Czech-English Translation
Petrželka, Jiří ; Schmidt, Marek (referee) ; Smrž, Pavel (advisor)
Tato diplomová práce popisuje principy statistického strojového překladu a demonstruje, jak sestavit systém pro statistický strojový překlad Moses. V přípravné fázi jsou prozkoumány volně dostupné bilingvní česko-anglické korpusy. Empirická analýza časové náročnosti vícevláknových nástrojů pro zarovnání slov demonstruje, že MGIZA++ může dosáhnout až pětinásobného zrychlení, zatímco PGIZA++ až osminásobného zrychlení (v porovnání s GIZA++). Jsou otestovány tři způsoby morfologického pre-processingu českých trénovacích dat za použití jednoduchých nefaktorových modelů. Zatímco jednoduchá lemmatizace může snížit BLEU, sofistikovanější přístupy většinou BLEU zvyšují. Positivní efekty morfologického pre-processingu se vytrácejí s růstem velikosti korpusu. Vztah mezi dalšími charakteristikami korpusu (velikost, žánr, další data) a výsledným BLEU je empiricky měřen. Koncový systém je natrénován na korpusu CzEng 0.9 a vyhodnocen na testovacím vzorku z workshopu WMT 2010.
|
|
|
Automatic Maneuver Identification from Flight Data Records
Mořkovský, Vít ; Vlk, Jan (referee) ; Chudý, Peter (advisor)
The aim of the bachelor thesis in the identification of maneuvers from flight data records. The flight with identified maneuvers is displayed in the space over a map base using the created web application. Initially, a set of rules to identify respective maneuvers was created. After that, the identification of maneuvers was implemented using the technique of cluster analysis (K-means) and a classification technique based on the Hidden Markov Model. The maneuvers identified using the Hidden Markov Model correspond to maneuvers identified using rules in 95.6 %.
|
|
|
Modern methods for protein secondary structure prediction and their comparison
Kraus, Ondřej ; Novotný, Marian (advisor) ; Pleskot, Roman (referee)
Today, there are several protein secondary structure predictors; most of them use algorithms such as hidden Markov models or artificial neural networks. Therefore I will introduce them to a reader in my thesis. I will explain their principles, as well as their advantages and disadvantages. The majority of contemporary predictors have accuracy 70%-80% for prediction of three types of protein secondary structure. However these results are only approximate, due to different testing methodology. Therefore the user should get familiar with the method and its testing methodology in detail at first. Key-words: protein structure prediction, hidden Markov model, artificial neural network, nearest neighbour, protein secondary structure
|
|
|
Dynamic Programming in Graph Algorithms
Biloš, Martin ; Křivka, Zbyněk (referee) ; Burgetová, Ivana (advisor)
This work is about graph algorithms, their use and the benefit of the optimization method of dynamic programming. This benefit is show to the user via the graphic application. Graph algorithms find use in many sectors of human activity. They are used in packet routing or, for example, navigation. There are three methods of graph algorithms used in this work. This problems are solved with classic and dynamic way and measured data are compared.
|
|
|
Analýza indexů akciových trhů a režimů na komoditních trzích
Kuchina, Elena ; Cahlík, Tomáš (advisor) ; Máša, Petr (referee) ; Lukáčik, Martin (referee)
The thesis focuses on the identification of the typical scenarios of the mutual relations among the stock markets considering different regimes on the commodity markets. For the identified scenarios the investment recommendations have been suggested. Considering different regimes the commodity markets go through and the mutual linkage among the stock markets during different situations on the commodity markets, six scenarios of the stock markets' mutual relations have been analyzed. It was shown that during most unstable period, when highly volatile regime prevails simultaneously on the energy, precious metals and non-energy commodity markets, the whole economy becomes to be more tied: the stock market indices demonstrate stronger interdependence, and as a consequence the benefits of diversification begin to fail. During the simultaneous presence of low volatility on all three analyzed commodity markets the agreement between occurrences of highly volatile state of most stock markets, besides the indices within the European region (DAX, CAC 40, IBEX 35), is rather weak. Similarly the correlation within regions and with other regions is weaker comparing with other situations on the commodity markets, so the standard investment strategy can be kept. It was also shown that the interdependence among the stock markets during the period of high volatility on the energy market differs depending on the source underlying the oil price shocks causing higher volatility. The regimes prevailing on the commodity and stock markets during different time periods have been detected by applying Hidden Markov Model methodology. To examine the similarity between the stock market indices in terms of highly volatile regimes' occurrences, Jaccard's similarity coefficient is employed. The correlation among the stock markets was computed by Spearman correlation coefficient. The final part of research is devoted to the model-based approach used to analyze the dependence of the movement direction of SSEC index on other stock market indices between two trading days during different situations on the commodity markets. The dependency analysis was performed by applying Stochastic Gradient Boosting methodology.
|
|
|
Machine Learning Concepts for Categorization of Objects in Images
Hubený, Marek ; Honec, Peter (referee) ; Horák, Karel (advisor)
This work is focused on objects and scenes recognition using machine learning and computer vision tools. Before the solution of this problem has been studied basic phases of the machine learning concept and statistical models with accent on their division into discriminative and generative method. Further, the Bag-of-words method and its modification have been investigated and described. In the practical part of this work, the implementation of the Bag-of-words method with the SVM classifier was created in the Matlab environment and the model was tested on various sets of publicly available images.
|
|
|
Modern methods for protein secondary structure prediction and their comparison
Kraus, Ondřej ; Novotný, Marian (advisor) ; Pleskot, Roman (referee)
Today, there are several protein secondary structure predictors; most of them use algorithms such as hidden Markov models or artificial neural networks. Therefore I will introduce them to a reader in my thesis. I will explain their principles, as well as their advantages and disadvantages. The majority of contemporary predictors have accuracy 70%-80% for prediction of three types of protein secondary structure. However these results are only approximate, due to different testing methodology. Therefore the user should get familiar with the method and its testing methodology in detail at first. Key-words: protein structure prediction, hidden Markov model, artificial neural network, nearest neighbour, protein secondary structure
|
|
|
Czech-English Translation
Petrželka, Jiří ; Schmidt, Marek (referee) ; Smrž, Pavel (advisor)
Tato diplomová práce popisuje principy statistického strojového překladu a demonstruje, jak sestavit systém pro statistický strojový překlad Moses. V přípravné fázi jsou prozkoumány volně dostupné bilingvní česko-anglické korpusy. Empirická analýza časové náročnosti vícevláknových nástrojů pro zarovnání slov demonstruje, že MGIZA++ může dosáhnout až pětinásobného zrychlení, zatímco PGIZA++ až osminásobného zrychlení (v porovnání s GIZA++). Jsou otestovány tři způsoby morfologického pre-processingu českých trénovacích dat za použití jednoduchých nefaktorových modelů. Zatímco jednoduchá lemmatizace může snížit BLEU, sofistikovanější přístupy většinou BLEU zvyšují. Positivní efekty morfologického pre-processingu se vytrácejí s růstem velikosti korpusu. Vztah mezi dalšími charakteristikami korpusu (velikost, žánr, další data) a výsledným BLEU je empiricky měřen. Koncový systém je natrénován na korpusu CzEng 0.9 a vyhodnocen na testovacím vzorku z workshopu WMT 2010.
|
guest ::
