This paper focuses on analysing and modelling inflation and its dynamics in Czech Republic applying a special kind of econometric models. Firstly economic theory of inflation is mentioned - fundamental terms, measuring methods of inflation, the way Czech national bank is monitoring the inflation and obviously a short summary of historical evolution of inflation in Czech economy. In the second part of this paper two econometric concepts of modelling time series are introduced - vector autoregression models (VAR models) and volatility models, concretely ARCH and GARCH models. In connection with the VAR models, Granger causality, impulse response functions, cointegration and error correction models are described. The empirical part includes application of selected models on real time series of chosen macroeconomic indicators. The estimation outputs are interpreted and forecasts are implemented. The quality of chosen econometric models for modelling inflation in Czech Republic is discussed.

L'activité de notre cerveau est intrinsèquement rythmique : des oscillations sont observées à tous les niveaux de son organisation. Cette rythmicité de l'activité cérébrale influence notre perception. En effet, au lieu de superviser continuellement notre environnement, notre cerveau effectue de brèves " clichés " du monde extérieur (entre 5 et 15 par seconde). Cela crée des cycles perpétuels : notre perception visuelle fluctue en fonction de la phase de l'oscillation sous- jacente. De nombreuses données témoignent du fait que les oscillations cérébrales à différentes fréquences sont fondamentales à la formation de notre perception visuelle. Lors de cette thèse, nous avons utilisé le Paradigme de Bruit Blanc comme outil pour comprendre l'influence des oscillations sur la perception visuelle et qui par extension pourra être utilisé pour contrôler cette perception. Le paradigme de bruit blanc visuel utilise des séquences de flashs dont la luminance varie aléatoirement (créant ainsi du " bruit blanc "), comme stimuli, qui contraignent l'activité cérébrale de manière prédictible. Les réponses impulsionnelles à ces séquences de bruit blanc sont caractérisées par une composante oscillatoire forte dans la bande alpha (~10Hz), similaire à un écho perceptuel. Puisque les réponses impulsionnelles sont un modèle de la réponse de notre cerveau à un flash dans la séquence de bruit blanc, elles peuvent être utilisées pour reconstruire (plutôt qu'enregistrer) l'activité cérébrale en réponse à de nouvelles séquences de stimulation. Par ailleurs, des cibles ont été introduites au sein des séquences de bruit blanc à un niveau proche du seuil de perception, et le décours temporel de cette activité reconstruite autour de la présentation des cibles a été extrait. Ainsi, l'EEG reconstruit peut être utilisé pour étudier l'influence de ces oscillations contraintes sur la perception visuelle, indépendamment des autres types de signaux généralement enregistrés dans l'EEG. Dans un premier temps, nous avons validé le paradigme de bruit blanc en montrant que : 1) les séquences de bruits blancs influencent bien la détection des cibles, 2) les échos perceptuels évoqués par les séquences de bruit blancs sont stables dans le temps, 3) ces échos sont un bon modèle de l'activité cérébrale enregistrée par EEG, et 4) leurs bases neuronales se situent dans les aires visuelles primaires. Dans un second temps, nous avons étudié la relation entre ces oscillations cérébrales contrôlées par la séquence de bruit blanc et la détection des cibles. Ici, nous montrons que l'activité EEG reconstruite nous aide à déterminer la véritable latence à laquelle la phase de l'oscillation (thêta) influence la perception. De plus, nous avons aussi montré que l'amplitude de l'oscillation (alpha) influence la détection des cibles et ce, indépendamment des fluctuations des facteurs endogènes (tel que l'attention). Enfin, tirant parti de ce lien entre oscillation et perception, nous construisons deux algorithmes qui permettent de contrôler la perception des sujets. Tout d'abord, nous mettons au point un modèle " universel " de la perception qui permet de prédire, pour n'importe quel observateur, si une cible dans une séquence de bruit blanc sera vue ou non. Ensuite, nous construisons un modèle individuel qui utilise l'écho perceptuel de chaque sujet comme clé de cryptage et nous permet de présenter des cibles à des moments où la cible sera détectée par un sujet seulement au détriment de tous les autres sujets, créant ainsi une sorte de système de cryptage neuronal (" Neuro-Encryption "). / Our brain activity is inherently rhythmic: oscillations can be found at all levels of organization. This rhythmicity in brain activity gives a rhythm to what we see: instead of continuously monitoring the environment, our brains take "snapshots" of the external world from 5 to 15 times a second. This creates perceptual cycles: depending on the phase of the underlying oscillation, our perceptual abilities fluctuate. Accumulating evidence shows that brains oscillations at various frequencies are instrumental in shaping visual perception. At the heart of this thesis lies the White Noise Paradigm, which we designed as a tool to better understand the influence of oscillations on visual perception and which ultimately could be used to control visual perception. The White Noise Paradigm uses streams of flashes with random luminance (i.e. white noise) as stimuli, which have been shown to constrain brain oscillations in a predictable manner. The impulse response to WN sequences has a strong (subject specific) oscillatory component at ~10Hz akin to a perceptual echo. Since the impulse response is a model of how our brains respond to one single flash in the sequence, they can be used to reconstruct (rather than record) the brain activity to new stimulation sequences. We then present near-perceptual threshold targets embedded within the WN sequences and extract the time course of these predicted/reconstructed background oscillations around target presentation. Thus, the reconstructed EEG can be used to study the influence of the oscillatory components on visual perception, independently of other types of signals usually recorded in the EEG. First, we validate the White Noise Paradigm by showing that: 1) the WN sequences do modulate behaviour, 2) the perceptual echoes evoked by these WN sequences are stable in time, 3) they are a (relatively) good model of the subject's recorded brain activity and 4) their neuronal basis can be found in the early visual areas. Second, we investigate the relationship between these constrained brain oscillations and visual perception. Specifically, we show that the reconstructed EEG can help us recover the true latency at which (theta) phase influences perception. Moreover, it can help us uncover a causal influence of (alpha) power on target detection, independently from any fluctuation in endogenous factors. Finally, capitalizing on the link between oscillations and perception, we build two algorithms used to control the perception of subjects. First, we build a "universal" forward model which can predict for any observer whether a particular target will be seen or not. Second, we build a subject-dependent model which can predict whether a particular subject (for whom EEG was recorded previously) will perceive a given target or not. Critically, this can be used to present targets optimized to be perceived by one subject only, to the detriment of all other subjects, creating a sort of "Neuro-Encryption" system.

RESUMO No início de 1995 foi adotado no Brasil o Plano Real, tendo como um dos seus tripés de sustentação a busca pelo combate ao processo inflacionário crônico brasileiro que já se estendia por um longo período. Assim, a política monetária passou a ter um papel importante na determinação das variáveis macroeconômicas. Este trabalho busca analisar uma regra de política monetária que capte as variações ocorridas em todo o período do Plano Real, se estendendo até meados de 2005, bem como se deram as relações entre as variáveis econômicas neste período. A especificação proposta consiste na estimação de modelos não-lineares distintos dependendo do estado da economia (em crise ou fora de crise). Utilizamos um modelo com chaveamento Markoviano para a dinâmica da taxa de juros nominal onde a determinação de períodos de crise é feita por uma variável nãoobservada. Além disso, procuramos adotar dois algoritmos distintos de estimação, Expectation-Maximization (EM) e Monte Carlo Markov Chain (MCMC), concluindo que a análise para ambos é bastante próxima, sendo identificados os mesmos períodos entre regimes. Finalmente, motivamos a estimação através de modelos econômicos teóricos cujas dinâmicas são compatíveis com uma regra de fixação de juros não-linear, avaliando os padrões de resposta a impulso condicionados ao estado da economia (regimes de estabilidade e crise econômica). / ABSTRACT In the beginning of 1995, continuing the process of inflation combat, the monetary policy should have been an important role in the determinacy of macroeconomics variables. This work has a target analyzing a monetary rule that reflects the occurred variations in every Real Plan?s period. The specification proposed by the authors consists in an estimation of two independent nonlinear models for different states of the nature (crises or not crises). Here we estimate a model where the dynamic of the nominal interest rate follows a Markov Switching process and the regimes are unobservable variables. In addition, we try adopting two different algorithms to estimation; Expectation-Maximization (EM) and Monte Carlo Markov Chain (MCMC), concluded that the results are very similar. Finally, we motivate the estimations analyzing models where the theoretical dynamics of the economy are compatible with a nonlinear interest rate rule, analyzing the impulse response conditioned to state of economy (regimes of crises or not crises).

<p>A wave digital filter is derived from an analog filter, which is realized as classical doubly resistively terminated reactancefilters. Perfectly designed wave digital filters express good dynamic signal range, low roundoff noise and excellent stabilitycharacteristics with respect to nonlinearity which are produced due to finite wordlength effects. Wave digital filters inheritthe sensitivity properties from analog filters, therefore, coefficients values can be selected to favorable values.Wave digital filters, derived from ladder filters, have low coefficient sensitivity in the passband and stopband. These WDFsare very complicated and are non-modular. The lattice wave digital filters are modular and are not complex. However, theyhave very high sensitivity in the stopband and thus require large coefficient wordlengths. The number of coefficients equalsthe filter order which have to be odd.This thesis discusses the wave digital filter structures that are modular because they are designed by cascading the first-orderand second-order sections. These WDFs can be pipelined. They also exhibit all the above mentioned favorable properties.Similar to lattice WDFs, these structures are restricted to symmetrical and antisymmetrical transfer functions. The synthesisof these structures is based on the factorization of the scattering matrix of lossless two-ports.In this thesis work, lowpass wave digital filters based on circulator-tree structure are designed with different orders startingfrom 3 and going upto 13. In parallel to these circulator-tree wave digital filters, the simple digital filters are also designedwith the same specification. The results of the two filters are compared with each other. It is observed that impulse responseand attenuation response of the two kind of filters perfectly match. Therefore, it is can be concluded that circulator-tree WDFupto Nth order can be synthesized. The implementation examples of two filter with order 3 and order 7 is presented in thisdocumentation for ready reference. It has also been shown that the order of sections does not affect the transfer function ofthe filter. Noise has been introduced and adaptor sections are penetrated. From the results it is concluded that the order of theadaptor sections does not matter and also that the noise does not affect the other adaptors sections, it only propagates throughother adaptors sections.</p>

A wave digital filter is derived from an analog filter, which is realized as classical doubly resistively terminated reactancefilters. Perfectly designed wave digital filters express good dynamic signal range, low roundoff noise and excellent stabilitycharacteristics with respect to nonlinearity which are produced due to finite wordlength effects. Wave digital filters inheritthe sensitivity properties from analog filters, therefore, coefficients values can be selected to favorable values.Wave digital filters, derived from ladder filters, have low coefficient sensitivity in the passband and stopband. These WDFsare very complicated and are non-modular. The lattice wave digital filters are modular and are not complex. However, theyhave very high sensitivity in the stopband and thus require large coefficient wordlengths. The number of coefficients equalsthe filter order which have to be odd.This thesis discusses the wave digital filter structures that are modular because they are designed by cascading the first-orderand second-order sections. These WDFs can be pipelined. They also exhibit all the above mentioned favorable properties.Similar to lattice WDFs, these structures are restricted to symmetrical and antisymmetrical transfer functions. The synthesisof these structures is based on the factorization of the scattering matrix of lossless two-ports.In this thesis work, lowpass wave digital filters based on circulator-tree structure are designed with different orders startingfrom 3 and going upto 13. In parallel to these circulator-tree wave digital filters, the simple digital filters are also designedwith the same specification. The results of the two filters are compared with each other. It is observed that impulse responseand attenuation response of the two kind of filters perfectly match. Therefore, it is can be concluded that circulator-tree WDFupto Nth order can be synthesized. The implementation examples of two filter with order 3 and order 7 is presented in thisdocumentation for ready reference. It has also been shown that the order of sections does not affect the transfer function ofthe filter. Noise has been introduced and adaptor sections are penetrated. From the results it is concluded that the order of theadaptor sections does not matter and also that the noise does not affect the other adaptors sections, it only propagates throughother adaptors sections.

In this thesis study, joint frequency offset and channel estimation methods for single-input single-output (SISO) systems are examined. The performance of maximum likelihood estimate of the parameters are studied for different training sequences. Conventionally training sequences are designed solely for the channel estimation purpose. We present a numerical comparison of different training sequences for the joint estimation problem. The performance comparisons are made in terms of mean square estimation error (MSE) versus SNR and MSE versus the total training energy metrics. A novel estimation scheme using complementary sequences have been proposed and compared with existing schemes. The proposed scheme presents a lower estimation error than the others in almost all numerical simulations. The thesis also includes an extension for the joint channel-frequency offset estimation problem to the multi-input multi-output systems and a brief discussion for multiple frequency offset case is also given.

Traditional naval vessels with steel structures have the benefit of large safety factors and a distinct material endurance limit. However, as performance requirements and budget constraints rise, the demand for lighter weight vessels increases. Reducing the mass of vessels is commonly achieved by the use of aluminum or composite structures, which requires closer attention to be paid to crack initiation and propagation. It is rarely feasible to require a lengthy inspection process that removes the vessel from service for an extended amount of time. Structural health monitoring (SHM), involving continuous measurement of the structural response to an energy source, has been proposed as a step towards condition-based maintenance. Furthermore, using a passive monitoring system with an array of sensors has several advantages: monitoring can take place in real-time using only ambient noise vibrations and neither deployment of an active source nor visual access to the inspected areas are required. Passive SHM on a naval vessel is not without challenge. The structures of ships are typically geometrically complex, causing scattering, multiple reflections, and mode conversion of the propagating waves in the vessel. And rather than a distinct and predictable input produced by controlled active sources, the vibration sources are hull impacts, smaller waves, and even onboard machinery and activity. This research summarizes findings from data collected onboard a Navy vessel and presents recommendations data processing techniques. The intent is to present a robust method of passive structural health monitoring for such a vessel using only ambient vibrations recordings.

System identification finds nowadays application in various areas of biological research as a tool of empiric mathematical modeling and model individualization. A fundamental challenge of system identification in biology awaits in the form of response variability. Furthermore, biological systems tend to exhibit high degree of nonlinearity as well as significant time delays. This thesis covers system identification approaches developed for the applications within two particular biomedical fields: neuroscience and endocrinology. The first topic of the thesis is parameter estimation of the classical Elementary Motion Detector (EMD) model in insect vision. There are two important aspects to be taken care of in the identification approach, namely the nonlinear dynamics of the individual EMD and the spatially distributed structure of multiple detectors producing a measurable neural response. Hence, the suggested identification method is comprised of two consecutive stages addressing each of the above aspects. Furthermore, visual stimulus design for high spatial excitation order has been investigated. The second topic is parameter estimation of mathematical model for testosterone regulation in the human male. The main challenges of this application are in the unavailability of input signal measurements and the presence of an unknown pulsatile feedback in the system resulting in a highly nonlinear closed-loop dynamics. Semi-blind identification method has been developed based on a recently proposed pulse-modulated model of pulsatile endocrine regulation. The two system identification problems treated in the thesis bear some resemblance in the sense that both involve measured signals that can be seen as square-integrable functions of time. This property is handled by transforming the signals into the Laguerre domain, i.e. by equivalently representing the functions with their infinite Laguerre series.

Esse trabalho está baseado na investigação dos detectores de falhas aplicando classificadores de classe única. As falhas a serem detectadas são relativas ao estado de funcionamento de cada componente do circuito, especificamente de suas tolerâncias (falha paramétrica). Usando a função de transferência de cada um dos circuitos são gerados e analisados os sinais de saída com os componentes dentro e fora da tolerância. Uma função degrau é aplicada à entrada do circuito, o sinal de saída desse circuito passa por uma função diferenciadora e um filtro. O sinal de saída do filtro passa por um processo de redução de atributos e finalmente, o sinal segue simultaneamente para os classificadores multiclasse e classe única. Na análise são empregados ferramentas de reconhecimento de padrões e de classificação de classe única. Os classficadores multiclasse são capazes de classificar o sinal de saída do circuito em uma das classes de falha para o qual foram treinados. Eles apresentam um bom desempenho quando as classes de falha não possuem superposição e quando eles não são apresentados a classes de falhas para os quais não foram treinados. Comitê de classificadores de classe única podem classificar o sinal de saída em uma ou mais classes de falha e também podem classificá-lo em nenhuma classe. Eles apresentam desempenho comparável ao classificador multiclasse, mas também são capazes detectar casos de sobreposição de classes de falhas e indicar situações de falhas para os quais não foram treinados (falhas desconhecidas). Os resultados obtidos nesse trabalho mostraram que os classificadores de classe única, além de ser compatível com o desempenho do classificador multiclasse quando não há sobreposição, também detectou todas as sobreposições existentes sugerindo as possíveis falhas. / This work deals with the application of one class classifiers in fault detection. The faults to be detected are related parametric faults. The transfer function of each circuit was generated and the outputs signals with the components in and out of tolerance were analyzed. Pattern recognition and one class classifications tools are employed to perform the analysis. The multiclass classifiers are able to classify the circuit output signal in one of the trained classes. They present a good performance when the fault classes do not overlap or when they are not presented to fault classes that were not presented in the training. The one class classifier committee may classify the output signal in one or more fault classes and may also classify them in none of the trained class faults. They present comparable performance to multiclass classifiers, but also are able to detect overlapping fault classes and show fault situations that were no present in the training (unknown faults).

Made available in DSpace on 2010-04-20T21:00:09Z (GMT). No. of bitstreams: 4 Alan Marinovic.pdf.jpg: 11803 bytes, checksum: 6f432bd77eda9bb1f8d0a1b55b2d1ea4 (MD5) Alan Marinovic.pdf.txt: 229382 bytes, checksum: 17f470747aee826fc7511c202d10bcc8 (MD5) Alan Marinovic.pdf: 1449291 bytes, checksum: bc20970ba9f69f9209cff4441caf931e (MD5) license.txt: 4886 bytes, checksum: 2dd7def8564dbf39bd3ed6b2e446baf6 (MD5) Previous issue date: 2009-01-28T00:00:00Z / O trabalho estuda a inter-relação entre preços de ações bancárias da América Latina, Estados Unidos e Europa durante o período compreendido entre janeiro de 2000 até final de junho de 2008. De um modo geral o estudo busca evidências sobre a existência de relações de equilíbrio de longo prazo entre as séries de preços utilizando análises de cointegração, testes de causalidade e funções de impulso resposta. Os resultados empíricos apontam para a existência de relações de equilíbrio de longo prazo entre as séries de preços, e para a existência de contágio especialmente de choques oriundos do mercado Norte Americano. Cabe ressaltar que o efeito de choques se mostra mais pronunciado após 2007, período compreendido pela crise do subprime. / This dissertation investigates the inter-relationships among bank’s stock markets for Latin America, United States and Europe from January, 2000 to June, 2008. The study analyzes the existence of long-run relationships among the price of Bank’s stocks, additionally it applies short-run causality tests and impulse response analyses. Empirical results suggests that there is at least one cointegration vector among the price series, and the series time paths are influenced by different extent of changes in price of different banks. Moreover the study finds that the relationships among bank’s stock prices differ between crises (more volatile) and less volatile periods. These findings imply that there are strong evidences of inter-connections among stock markets around the world.