|
|
Mining of Textual Data from the Web for Speech Recognition
Kubalík, Jakub ; Plchot, Oldřich (referee) ; Mikolov, Tomáš (advisor)
Prvotním cílem tohoto projektu bylo prostudovat problematiku jazykového modelování pro rozpoznávání řeči a techniky pro získávání textových dat z Webu. Text představuje základní techniky rozpoznávání řeči a detailněji popisuje jazykové modely založené na statistických metodách. Zvláště se práce zabývá kriterii pro vyhodnocení kvality jazykových modelů a systémů pro rozpoznávání řeči. Text dále popisuje modely a techniky dolování dat, zvláště vyhledávání informací. Dále jsou představeny problémy spojené se získávání dat z webu, a v kontrastu s tím je představen vyhledávač Google. Součástí projektu byl návrh a implementace systému pro získávání textu z webu, jehož detailnímu popisu je věnována náležitá pozornost. Nicméně, hlavním cílem práce bylo ověřit, zda data získaná z Webu mohou mít nějaký přínos pro rozpoznávání řeči. Popsané techniky se tak snaží najít optimální způsob, jak data získaná z Webu použít pro zlepšení ukázkových jazykových modelů, ale i modelů nasazených v reálných rozpoznávacích systémech.
|
|
|
Entity
Jindra, Jakub ; Turek, Filip (referee) ; Hájek, Václav (advisor)
Informace může vytvářet určité entropické krajiny. Entropická krajina je složená ze zdrojů informace. Zdroj může být informačně přínosný nebo úplně zbytečný. Dobrý zdroj vytváří nízkou entropii, vysoký náboj, a špatný zdroj entropii vysokou s nízkým informačním nábojem. Touto optikou pak můžeme vidět fiktivní krajinu, která se kolem nás nalézá. Entita je její vizualizací, vytváří útesy, tam kde byly roviny, hory zkoncentrované informace a měsíční údolí s nulovou hodnotou.
|
|
|
The Financial Secrecy Index: An Information Theory Approach
Galuszka, Lukáš ; Janský, Petr (advisor) ; Malovaná, Simona (referee)
The objective of this thesis is to evaluate alternative weighting systems to determine if they have the potential to improve the current weighting system of the Financial Secrecy Index (FSI). The FSI, a measure of countries' contributions to global financial secrecy, currently weights its 15 qualitative components equally. A web-based opinion survey conducted in January and February 2016 among academics, journalists, experts and other persons familiar with FSI serves as the baseline for assessing new weights. The new weights derived from the survey results are not significantly different from the equal weights in 14 out of 15 components. The survey results suggest that widely held opinion is consistent with equal weight assumptions. Statistical model selection criteria from information theory that penalize model complexity prefer in majority of cases the simple model over the more complex one even though more complex model provides better goodness-of-fit statistics. Alternative methods and analysis such as Principal Components Analysis is performed and discussed. The present work finds that, statistically, the weights should not diverge from the equal weighting system in use currently. Powered by TCPDF (www.tcpdf.org)
|
|
|
Diagnostic parameters determination of aircraft hydraulic system
Třetina, Karel ; Cäsar, Tomáš
The article presents methodology for application of information theory for selection of hydraulic system diagnostic parameters. Changes in technical state of the system are accompanied by a set of symptoms. Method of determining information value of each symptom by information theory is presented. A set of technical conditions and set of symptoms for a simple hydraulic system are determined. Matrix of technical state describing relations between technical conditions and symptoms is constructed. The four symptoms are determined by matrix analysis with use of information theory. These symptoms are chosen as diagnostic parameters and provide the most information about technical state of the hydraulic system.
|
|
|
Image, Information, Complexity. The study of visual information incorporating the function of information entropy with a focus on abstract art
Malečková, Dita ; Kera, Denisa (advisor) ; Marcelli, Miroslav (referee) ; Havel, Ivan M. (referee)
This text focuses on the relation of information and image, hence Information Theory and Image Analysis, as well as visualization of information and methods of visual analytics focusing on analysis of art works. It also concentrates on evolution of digital image and related new type of perception and artificial aesthetics. We narrow the broader topic of the image and image information to the abstract art, namely the work of Czech painter Frantisek Kupka, which is used as input in the experiment presenting original method of image analysis using the function of information entropy (Rényi entropy). This approach was used for the first time for analysis of art works with the aim to obtain the comparaison of natural and artificial classification of image information. We chose the work of abstract art not only with regard to given history of grammatics of abstract forms and its relation to the digital image, but also as an emblematic example of effective gaining of information from complex environment. Work thus summarizes historical context of evolution of digital image and theoretical reflection of contemporary image analytics and others techniques relevant to the image information and emphasizes relation of abstract art to the natural and simulated complexity.
|
|
|
Avoiding overfitting of models: an application to research data on the Internet videos
Jiroušek, Radim ; Krejčová, I.
The problem of overfitting is studied from the perspective of information theory. In this context, data-based model learning can be viewed as a transformation process, a process transforming the information contained in data into the information represented by a model. The overfitting of a model often occurs when one considers an unnecessarily complex model, which usually means that the considered model contains more information than the original data. Thus, using one of the basic laws of information theory saying that any transformation cannot increase the amount of information, we get the basic restriction laid on models constructed from data: A model is acceptable if it does not contain more information than the input data file.
|
|
|
The Financial Secrecy Index: An Information Theory Approach
Galuszka, Lukáš ; Janský, Petr (advisor) ; Malovaná, Simona (referee)
The objective of this thesis is to evaluate alternative weighting systems to determine if they have the potential to improve the current weighting system of the Financial Secrecy Index (FSI). The FSI, a measure of countries' contributions to global financial secrecy, currently weights its 15 qualitative components equally. A web-based opinion survey conducted in January and February 2016 among academics, journalists, experts and other persons familiar with FSI serves as the baseline for assessing new weights. The new weights derived from the survey results are not significantly different from the equal weights in 14 out of 15 components. The survey results suggest that widely held opinion is consistent with equal weight assumptions. Statistical model selection criteria from information theory that penalize model complexity prefer in majority of cases the simple model over the more complex one even though more complex model provides better goodness-of-fit statistics. Alternative methods and analysis such as Principal Components Analysis is performed and discussed. The present work finds that, statistically, the weights should not diverge from the equal weighting system in use currently. Powered by TCPDF (www.tcpdf.org)
|
|
|
Entropy and Consumerism in Pynchon's Works
Litochleb, Daniel ; Roraback, Erik Sherman (advisor) ; Robbins, David Lee (referee)
In his works, Pynchon engages the difficult concept of entropy and exposes how certain trends in contemporary culture and society, such as consumerism, exhibit tendencies similar to those of entropy. This observation has disturbing implications for contemporary culture and society, such as eventual cultural heat-death, where new ideas cease to develop, because all the useful intellectual energy has been exhausted. Therefore, I have decided to examine this concept of entropy and how Pynchon engages it in his works "Entropy," The Crying of Lot 49 and Gravity's Rainbow, which expose the main contributors to greater entropy and the disturbing implications of this concept for contemporary culture and society, some of which we are witnessing even today. The two main scientific types of entropy, thermodynamic and that of information theory, are explored by Pynchon in his early short story "Entropy," which I have analyzed in the second chapter of this thesis. As such, this short story provides an unobstructed view of Pynchon's early engagement with the concept of entropy, which increases in complexity in his later works The Crying of Lot 49 and Gravity's Rainbow. The Crying of Lot 49, explored in the third chapter of this thesis, indicates a considerable development of and change in Pynchon's perspective...
|
|
|
Theoretical predictions of the impact of a noisy channel on interpreters' performance in simultaneous interpretation and their empirical verification
Volenová, Eva ; Čeňková, Ivana (advisor) ; Ďoubalová, Jana (referee)
This thesis explores the issue of noise in the simultaneous interpreting process, using the concept of noise as defined in information theory and the model of communication developed by Claude Shannon. The theoretical section of this thesis first offers a brief explanation of the basic concepts involved in information theory, followed by a description of Shannon's model. Selected models of interpreting are then introduced based on this model of communication. A list of potential types of noise is then presented based on the literature examined that may have an impact on the interpreter during simultaneous interpreting. The empirical section of this thesis consists of a questionnaire and an experiment. The questionnaire determines to what extent interpreters find the potential types of noise from the list created in the theoretical section disruptive in their work. The list of types of noise is also expanded based on respondents' answers. The questionnaire also determines what strategies respondents employ to overcome adverse factors. The new types of noise and strategies employed are described in this thesis using open-ended answers from the respondents. The experiment tests the impact of a burst-noise channel on simultaneous interpreting. A total of 13 respondents took part in the experiment,...
|
|
|
Mining of Textual Data from the Web for Speech Recognition
Kubalík, Jakub ; Plchot, Oldřich (referee) ; Mikolov, Tomáš (advisor)
Prvotním cílem tohoto projektu bylo prostudovat problematiku jazykového modelování pro rozpoznávání řeči a techniky pro získávání textových dat z Webu. Text představuje základní techniky rozpoznávání řeči a detailněji popisuje jazykové modely založené na statistických metodách. Zvláště se práce zabývá kriterii pro vyhodnocení kvality jazykových modelů a systémů pro rozpoznávání řeči. Text dále popisuje modely a techniky dolování dat, zvláště vyhledávání informací. Dále jsou představeny problémy spojené se získávání dat z webu, a v kontrastu s tím je představen vyhledávač Google. Součástí projektu byl návrh a implementace systému pro získávání textu z webu, jehož detailnímu popisu je věnována náležitá pozornost. Nicméně, hlavním cílem práce bylo ověřit, zda data získaná z Webu mohou mít nějaký přínos pro rozpoznávání řeči. Popsané techniky se tak snaží najít optimální způsob, jak data získaná z Webu použít pro zlepšení ukázkových jazykových modelů, ale i modelů nasazených v reálných rozpoznávacích systémech.
|
guest ::
