A globalização, a maior ou menor estabilidade financeira e política e o atual nível de desenvolvimento econômico que o Brasil vem apresentando, condição que lhe concedeu a classificação de investment grade, obtido em 30 de abril de 2008, pela Agência Standard & Poor’s, conjugado com o nível de renda que a população vem atingindo, está transformando o país num mercado atrativo ao capital externo. A regulamentação do setor de saúde privada iniciada com a lei 9656/98 e com a criação da Agência Nacional de Saúde Suplementar – ANS, em 2000, fez com que este segmento mercadológico passasse a atrair a curiosidade e, agora, o direcionamento da atenção de mega investidores. Neste mercado há uma importante participação das cooperativas médicas e odontológicas - com significativa parte no share – aproximadamente 40%, no seu conjunto. A regulamentação vem causando certa concentração do setor pela aquisição de carteiras ou até mesmo de toda a empresa, sendo que estas negociações atingem a casa dos bilhões de reais. Essas situações encontram nas sociedades cooperativas médicas e odontológicas peculiaridades que divergem destas facilidades de negociação. / Globalization, financial and political stability and current degree of economical development that Brazil are living (because of these condition Standard & Poor’s gave to Brazil the classification of investment grade on April 30th, 2008) combined with Brazilians income degree are converting Brazil in an attractive market for outside capital. Private health sector regulamentation which started with the Brazilian law number 9656 from 1998, and the creation of Agência Nacional de Saúde Suplementar – ANS (in 2000) let this sector to attract curiosity and nowadays the attention of big stakeholders. Medical and odontological cooperatives have important participation in this market with signification part in share – like 40% on its set. The regulamentation bring on a concentration in the sector by portfolio purchase or until same all firm acquisition – that involves trillion of Brazilian reais. These situations find on medical and odontological cooperatives quirks that diverge from negotiations facilities.

Cette thèse décrit la modélisation et la simulation de systèmes à deux phases composées de particules évoluant dans un gaz. Les deux phases interagissent entre elles et le type de modèle à considérer dépend directement du type de simulations envisagées. Dans une première partie, les deux phases sont considérées comme des fluides, elles sont décrites à l’aide d’un modèle de mélange avec une relation de dérive (permettant de suivre une vitesse relative entre les deux phases et de prendre en compte deux vitesses) et sont supposées à l’équilibre en température et pression. Cette partie du manuscrit est composée de la dérivation des équations, de l’écriture d’un schéma numérique associé à ce jeu d’équations, d’une étude d’ordre de ce schéma ainsi que de simulations. Une étude mathématique de ce modèle (hyperbolicité dans un cadre simplifié, stabilité du système linéaire autour d’un état constant) a été réalisée dans un cadre o`u le gaz est supposé barotrope. La seconde partie de ce manuscrit est consacrée à la modélisation de l’effet de collisions inélastiques sur les gouttelettes lorsque l’on se place à un temps de simulation beaucoup plus court, pour lequel les gouttelettes ne peuvent plus être vues comme un fluide. Pour modéliser ces collisions, on construit un modèle simplifié (moins coûteux en temps) de type BGK permettant de reproduire le comportement en temps de certains moments sur les gouttelettes. Ces moments sont choisis pour être représentatifs de l’effet des collisions sur ces gouttelettes, à savoir une thermalisation en vitesse et énergie. Ce modèle est discrétisé avec une méthode particulaire et des résultats numériques sont donnés en comparaison avec ceux obtenus avec un modèle résolvant directement l’équation de Boltzmann homogène. / This thesis describes the modelisation and the simulation of two-phase systems composed of droplets moving in a gas. The two phases interact with each other and the type of model to consider directly depends on the type of simulations targeted. In the first part, the two phases are considered as fluid and are described using a mixture model with a drift relation (to be able to follow the relative velocity between the two phases and take into account two velocities), the two-phase flows are assumed at the equilibrium in temperature and pressure. This part of the manuscript consists of the derivation of the equations, writing a numerical scheme associated with this set of equations, a study of this scheme and simulations. A mathematical study of this model (hyperbolicity in a simplified framework, linear stability analysis of the system around a steady state) was conducted in a frame where the gas is assumed barotropic. The second part is devoted to the modelisation of the effect of inelastic collisions on the particles when the time of the simulation is shorter and the droplets can no longer be seen as a fluid. We introduce a model of inelastic collisions for droplets in a spray, leading to a specific Boltzmann kernel. Then, we build caricatures of this kernel of BGK type, in which the behavior of the first moments of the solution of the Boltzmann equation (that is mass, momentum, directional temperatures, variance of the internal energy) are mimicked. The quality of these caricatures is tested numerically at the end.

In 3D modelling software, deformations are used to add, to remove, or to modify geometric features of existing 3D models to create new models with similar but slightly different details. Traditional techniques for deforming virtual 3D models require users to explicitly define control points and regions of interest (ROIs), and to define precisely how to deform ROIs using control points. The awkwardness of defining these factors in traditional 3D modelling software makes it difficult for people with limited experience of 3D modelling to deform existing 3D models as they expect. As applications which require virtual 3D model processing become more and more widespread, it becomes increasingly desirable to lower the "difficulty of use" threshold of 3D model deformations for users. This thesis argues that the user experience, in terms of intuitiveness and ease of use, of a user interface for deforming virtual 3D models, can be greatly enhanced by employing sketch-based 3D model deformation techniques, which require the minimal quantities of interactions, while keeping the plausibility of the results of deformations as well as the responsiveness of the algorithms, based on modern home grade computing devices. A prototype system for sketch-based 3D model deformations is developed and implemented to support this hypothesis, which allows the user to perform a deformation using a single deforming stroke, eliminating the need to explicitly select control points, the ROI and the deforming operation. GPU based accelerations have been employed to optimise the runtime performance of the system, so that the system is responsive enough for real-time interactions. The studies of the runtime performance and the usability of the prototype system are conducted to provide evidence to support the hypothesis.

<p>Artificial neural networks, including deep neural networks, play a central role in data-driven science due to their superior learning capacity and adaptability to different tasks and data structures. However, although quantitative uncertainty analysis is essential for training and deploying reliable data-driven models, the uncertainties in neural networks are often overlooked or underestimated in many studies, mainly due to the lack of a high-fidelity and computationally efficient uncertainty quantification approach. In this work, a novel uncertainty analysis scheme is developed. The Gaussian mixture model is used to characterize the probability distributions of uncertainties in arbitrary forms, which yields higher fidelity than the presumed distribution forms, like Gaussian, when the underlying uncertainty is multimodal, and is more compact and efficient than large-scale Monte Carlo sampling. The fidelity of the Gaussian mixture is refined through adaptive scheduling of the width of each Gaussian component based on the active assessment of the factors that could deteriorate the uncertainty representation quality, such as the nonlinearity of activation functions in the neural network. </p> <p>Following this idea, an adaptive Gaussian mixture scheme of nonlinear uncertainty propagation is proposed to effectively propagate the probability distributions of uncertainties through layers in deep neural networks or through time in recurrent neural networks. An adaptive Gaussian mixture filter (AGMF) is then designed based on this uncertainty propagation scheme. By approximating the dynamics of a highly nonlinear system with a feedforward neural network, the adaptive Gaussian mixture refinement is applied at both the state prediction and Bayesian update steps to closely track the distribution of unmeasurable states. As a result, this new AGMF exhibits state-of-the-art accuracy with a reasonable computational cost on highly nonlinear state estimation problems subject to high magnitudes of uncertainties. Next, a probabilistic neural network with Gaussian-mixture-distributed parameters (GM-PNN) is developed. The adaptive Gaussian mixture scheme is extended to refine intermediate layer states and ensure the fidelity of both linear and nonlinear transformations within the network so that the predictive distribution of output target can be inferred directly without sampling or approximation of integration. The derivatives of the loss function with respect to all the probabilistic parameters in this network are derived explicitly, and therefore, the GM-PNN can be easily trained with any backpropagation method to address practical data-driven problems subject to uncertainties.</p> <p>The GM-PNN is applied to two data-driven condition monitoring schemes of manufacturing processes. For tool wear monitoring in the turning process, a systematic feature normalization and selection scheme is proposed for the engineering of optimal feature sets extracted from sensor signals. The predictive tool wear models are established using two methods, one is a type-2 fuzzy network for interval-type uncertainty quantification and the other is the GM-PNN for probabilistic uncertainty quantification. For porosity monitoring in laser additive manufacturing processes, convolutional neural network (CNN) is used to directly learn patterns from melt-pool patterns to predict porosity. The classical CNN models without consideration of uncertainty are compared with the CNN models in which GM-PNN is embedded as an uncertainty quantification module. For both monitoring schemes, experimental results show that the GM-PNN not only achieves higher prediction accuracies of process conditions than the classical models but also provides more effective uncertainty quantification to facilitate the process-level decision-making in the manufacturing environment.</p><p>Based on the developed uncertainty analysis methods and their proven successes in practical applications, some directions for future studies are suggested. Closed-loop control systems may be synthesized by combining the AGMF with data-driven controller design. The AGMF can also be extended from a state estimator to the parameter estimation problems in data-driven models. In addition, the GM-PNN scheme may be expanded to directly build more complicated models like convolutional or recurrent neural networks.</p>

This thesis deals with an autonomous single-channel deinterleaving. An autonomous single-channel deinterleaving is a separation of the received sequence of impulses from more than one emitter to sequences of impulses from one emitter without a human assistance. Methods used for deinterleaving could be divided into single-parameter and multiple-parameter methods according to the number of parameters used for separation. This thesis primarily deals with multi-parameter methods. As appropriate methods for an autonomous single-channel deinterleaving DBSCAN and variational bayes methods were chosen. Selected methods were adjusted for deinterleaving and implemented in programming language Python. Their efficiency is examined on simulated and real data.

Diplomová práce se zabývá porozuměním dynamických scén pro navigaci mobilních robotů. V první části předkládáme nový přístup k "sebe-učícím" modelům - fůzi odhadu úběžníku cesty založeného na frekvenčním zpracování a pravděpodobnostních modelech využívající barvu pro segmentaci. Detekce úběžníku cesty je založena na odhadu dominantních orientací texturního toku, získáného pomocí banky Gaborových vlnek, a hlasování. Úběžník cesty poté definuje trénovací oblast, která se využívá k samostatnému učení barevných modelů. Nakonec, oblasti tvořící cestu jsou vybrány pomocí měření Mahalanobisovi vzdálenosti. Pár pravidel řeší situace, jako jsou mohutné stíny, přepaly a rychlost adaptivity. Kromě toho celý odhad úběžníku cesty je přepracován - vlnky jsou nahrazeny aproximacemi pomocí binárních blokových funkcí, což umožňuje efektivní filtraci pomocí integrálních obrazů. Nejužší hrdlo celého algoritmu bylo samotné hlasování, proto překládáme schéma, které nejdříve provede hrubý odhad úběžníku a následně jej zpřesní, čímž dosáhneme výrazně vyšší rychlosti (až 40x), zatímco přesnost se zhorší pouze o 3-5%. V druhé části práce předkládáme vyhlazovací filtr pro prostorovo-časovou konzistentnost predikcí, která je důležitá pro vyspělé systémy. Klíčovou částí filtru je nová metrika měřící podobnost mezi třídami, která rozlišuje mnohem lépe než standardní Euclidovská vzdálenost. Tato metrika může být použita k nejrůznějším úlohám v počítačovém vidění. Vyhlazovací filtr nejdříve odhadne optický tok, aby definoval lokální okolí. Toto okolí je použito k rekurzivní filtraci založené na podobnostní metrice. Celková přesnost předkládané metody měřená na pixelech, které nemají shodné predikce mezi původními daty a vyfiltrovanými, je téměř o 18% vyšší než u původních predikcí. Ačkoliv využíváme SHIM jako zdroj původních predikcí, algoritmus může být kombinován s kterýmkoliv jiným systémem (MRF, CRF,...), který poskytne predikce ve formě pravěpodobností. Předkládaný filtr představuje první krok na cestě k úplnému usuzování.

This thesis is focused on knowledge discovery from databases, especially on methods of classification and prediction. These methods are described in detail.  Furthermore, this work deals with multimedia databases and the way these databases store data. In particular, the method for processing low-level image and video data is described.  The practical part of the thesis focuses on the implementation of this GMM method used for extracting low-level features of video data and images. In other parts, input data and tools, which the implemented method was compared with, are described.  The last section focuses on experiments comparing extraction efficiency features of high-level attributes of low-level data and the methods implemented in selected classification tools LibSVM.

Tato práce se zabývá pravděpodobnostními modely pro automatické rozpoznávání řečníka. Podrobně analyzuje zejména pravděpodobnostní lineární diskriminační analýzu (PLDA), která modeluje nízkodimenzionální reprezentace promluv ve formě \acronym{i--vektorů}.  Práce navrhuje dvě rozšíření v současnosti požívaného PLDA modelu. Nově navržený PLDA model s plným posteriorním rozložením  modeluje neurčitost při generování i--vektorů. Práce také navrhuje nový diskriminativní přístup k trénování systému pro verifikaci řečníka, který je založený na PLDA. Pokud srovnáváme původní PLDA s modelem rozšířeným o modelování  neurčitosti i--vektorů, výsledky dosažené s rozšířeným modelem dosahují až 20% relativního zlepšení při testech s krátkými nahrávkami. Pro delší  testovací segmenty  (více než jedna minuta) je zisk v přesnosti  menší, nicméně přesnost nového modelu není nikdy menší než přesnost výchozího systému.  Trénovací data jsou ale obvykle dostupná ve formě dostatečně dlouhých segmentů, proto v těchto případech použití nového modelu neposkytuje žádné výhody při trénování. Při trénování může být použit původní PLDA model a jeho rozšířená verze může být využita pro získání skóre v  případě, kdy se bude provádět testování na krátkých segmentech řeči. Diskriminativní model je založen na klasifikaci dvojic i--vektorů do dvou tříd představujících oprávněný a neoprávněný soud (target a non-target trial). Funkcionální forma pro získání skóre pro každý pár je odvozena z PLDA a trénování je založeno na logistické regresi, která minimalizuje vzájemnou entropii mezi správným označením všech soudů a pravděpodobnostním označením soudů, které navrhuje systém. Výsledky dosažené s diskriminativně trénovaným klasifikátorem jsou podobné výsledkům generativního PLDA, ale diskriminativní systém prokazuje schopnost produkovat lépe kalibrované skóre. Tato schopnost vede k lepší skutečné přesnosti na neviděné evaluační sadě, což je důležitá vlastnost pro reálné použití.

This thesis inquires the existing methods on the field of image recognition with regards to gesture recognition. Some methods have been chosen for deeper study and these are to be discussed later on. The second part goes in for the concenpt of an algorithm that would be able of robust gesture recognition based on data acquired within the AMI and M4 projects. A new ways to achieve precise information on participants position are suggested along with dynamic data processing approaches toward recognition. As an alternative, recognition using Gaussian Mixture Models and periodicity analysis are brought in. The gesture class in focus are speech supporting gestures. The last part demonstrates the results and discusses future work.

This master's thesis is dedicated to theme of knowledge discovery in Multimedia Databases, especially basic methods of classification and prediction used for data mining. The other part described about extraction of low level features from video data and images and summarizes information about content-based search in multimedia content and indexing this type of data. Final part is dedicated to implementation Gaussian mixtures model for classification and compare the final result with other method SVM.