A Educação Baseada em Web vem crescendo substancialmente ao longo dos últimos anos; em decorrência desse fator, aumenta a diversidade de alunos que buscam através desse meio uma alternativa para sua formação, seja ela acadêmica ou profissional. Esses alunos possuem diferentes características pessoais, sócio-culturais e cognitivas, e desta maneira possuem preferências e necessidades variadas. Nesse contexto, os Sistemas Hipermídia Adaptativos oferecem um recurso para atender a esse público diverso, que tem um objetivo em comum, o ensino a distância por meio da Web. O modelo aqui proposto versa sobre um Sistema Hipermídia Adaptativo, cujos critérios para adaptação consistem na seleção de conteúdos apropriados ao estilo cognitivo do aluno, o que é identificado através da investigação das suas Inteligências Múltiplas. Para isso, foram propostas 10 personas-cognitivas, resultantes da combinação das inteligências mais desenvolvidas em cada um. Também foi realizado um estudo sobre a relação entre as Inteligências Múltiplas e a Semiótica e entre o conteúdo instrucional e a Semiótica. Desse estudo foi originada uma matriz, resumindo os resultados obtidos e que foi utilizada como alicerce para a seleção dos signos mais adequados a cada uma das personas. Utilizou-se como Modelo de Referência para Hipermídia Adaptativa o Modelo Munich, que foi adequado à proposta educacional e ao estudo dos signos. Foi então desenvolvido o modelo SHASIM, onde se contemplam os principais aspectos para concepção de um Sistema Hipermídia Adaptativo para o domínio educacional. / The Web Based Education have grown substantially along the last years; in consequence of this factor, it increases the student\'s diversity that seeks through this midia an alternative for their formation, being it academic or professional. These students possesses different personal, social-cultural and cognitive characteristics, and in this way, they possesses various preferences and needs. In that context, the Adaptative Hipermidia Sistems offers resources to assist that various public, that has one common objective, the distance teaching through the Web. The model here proposed is about a Adaptative Hipermidia System, whose criteria for adaptation consists of the appropriate contents selection for the student\'s cognitive style, that is identified through the investigation of their Multiple Intelligences. For that were proposed 10 cognitive-personas, resultants of the combination of the most developed intelligences in each one. Also, a study was accomplished about the relationship between the Multiple Intelligences and the Semiotics and also between the instructional content and the Semiotics. From this study was originated a matrice, summarizing the obtained results and it was used as a foundation for the selection of the most appropriate signs to each one of the personas. It was used as the Adaptative Hipermidia Reference Model the Munich Model, that was adapted to the education proposal and signs study. It was then developed the SHASIM model, where are contemplated the main aspects for conception of a Adaptative Hipermidia System for the education domain.

A busca por substâncias imunomoduladoras oriundas de plantas medicinais é crescente devido à grande variedade de doenças relacionadas às respostas imunológicas desordenadas, como as autoimunidades e o câncer. Neste sentido, sabe-se que a Paullinia cupana, popularmente conhecida como guaraná, é uma planta medicinal que apresenta entre outros efeitos, propriedades quimiopreventiva e antitumoral em camundongos. Entretanto, até o presente estudo, não havia relatos na literatura a respeito dos efeitos desta planta sobre o sistema imune, conhecido por desempenhar papel fundamental não só no controle de infecções, mas também no combate ao câncer. Assim, o presente estudo avaliou os efeitos do pó das sementes de guaraná e de seus principais compostos ativos, cafeína e catequina, sobre o sistema imune de camundongos C57BL/6 através de protocolos preconizados pela Organization for Economic Cooperation and Development (OECD) com modificações. Os resultados obtidos mostraram que o guaraná, devido à ação majoritária da catequina, mas com a participação da cafeína também, diminui as respostas imunes, celular e humoral, verificada através da diminuição da expansão de linfócitos T frente a um antígeno específico, redução da resposta de hipersensibilidade do tipo tardia e título de anticorpos, diminuição dos pesos relativos do baço e timo e celularidade deste último, além de aumento de células imaturas no timo. Desta forma, os resultados aqui expostos demonstram, pela primeira vez, que a ingestão de guaraná reduz a imunidade adaptativa através da diminuição das respostas celular e humoral. Assim, futuros estudos deverão ser realizados, no sentido de se verificar a possibilidade do uso desta planta ou de seus princípios ativos no controle de doenças inflamatórias e autoimunes, caracterizadas por uma resposta exacerbada do sistema imune. / Seeking for immunomodulatory compounds derived from medicinal plants is increasing due to the huge variety of diseases related to disordered immune responses, such as autoimmunity and cancer. Therefore, it is known that the Paullinia cupana, popularly known as guarana, is a medicinal plant that has among other effects, chemopreventive and antitumor properties in mice. However, until the present moment, there were no reports in the literature about the effects of this plant on the immune system, which is known to play key role not only in the infectious control, but also in combat cancer. Thus, this study evaluated the effects of guarana seed powder and its major active principles, caffeine and catechin, on the immune system of C57BL/6 mice through protocols recommended by the Organization for Economic Cooperation and Development (OECD) with modifications. The results showed that guarana, due to the majority action of catechin, but also with the participation of caffeine, decreases the cellular and humoral immune responses, as indicated by the decreased expansion of antigen-specific T lymphocyte, reduced response delayed-type hypersensitivity and titer of antibodies, reduction in relative weights of thymus and spleen cellularity, besides increase of immature cells in the thymus. Thus, the results showed here demonstrate, for the first time, that the intake of guarana decreases the adaptive immunity by reducing cellular and humoral responses. Thus, futures studies should be conducted in order to better evaluate the use of this plant and/or its main active principle in the treatment of inflammatory and autoimmune diseases, which are characterized by exarcebated immune response.

Os jogos podem ser utilizados como uma forma mais interativa e lúdica de aprender. Porém, os alunos possuem preferências diferentes no aprendizado, denominados estilos de aprendizagem. Esses estilos indicam diferenças ao absorver, processar e transformar as informações transmitidas ao aluno em conhecimento. Um modelo de estilos de aprendizagem dedicado ao ensino de engenharia ´e o modelo Felder-Silverman, que possui 4 dimensões descrevendo as preferências dos alunos ao aprender. A proposta deste projeto ´e desenvolver um jogo educativo de física para alunos do curso de Engenharia, que consiga identificar estes estilos e alterar suas características para atender as necessidades dos diferentes jogadores, implementando um comportamento adaptativo, sem utilizar a ferramenta para a medição dos estilos criada por Felder. Inicialmente, foi feito um experimento de design centrado no usuário para descobrir as características de jogos existentes de física agradavam ou incomodavam os alunos do curso de engenharia. Depois disso, o jogo foi desenvolvido em Unity, abordando o tema de Atrito, cuja simulação física foi alterada para não incluir conceitos mais avançados. O jogo inicial falhou em atrair a atenção de jogadores na faixa etária do público alvo, então sua interface foi redesenhada para a temática de um laboratório de testes. Esta segunda versão foi testada com quatro alunos do curso de engenharia da computação do Insper e a adaptatividade não ocorreu em nenhuma das sessões de jogo. Além disso, utilizando uma ferramenta de avaliação de jogos adaptativos denominada MEEGA+, foi apontado que o jogo possuía uma interface de fácil aprendizado, mas que seus desafios eram muito fáceis e não criavam uma boa experiência para o jogador. Mais refinamento é necessário para o modelo da adaptatividade e para o design do jogo. / Games can be used as a more interactive and playful way of learning. However, students have different preferences in learning, called learning styles. These styles indicate differences in absorbing, processing, and transforming information passed on to the student in knowledge. A model of learning styles dedicated to engineering education is the Felder-Silverman model, which has 4 dimensions describing students\' preferences in learning. The proposal of this project is to develop an educational game of physics for students of the Engineering course, who can identify these styles and change their characteristics to meet the needs of different players, implementing an adaptive behavior, without using the tool for measuring the styles created by Felder. Initially, a user-centered design experiment was done to discover the characteristics of existing physics games that pleased or bothered students in the engineering course. After that, the game was developed in Unity, addressing the theme of Friction, whose physical simulation was changed to not include more advanced concepts. The initial game failed to attract the attention of players in the age range of the target audience, so its interface was redesigned to the subject of a test lab. This second version was tested with four students from Insper\'s computer engineering course and adaptivity did not occur in any of the game sessions. In addition, using an adaptive game evaluation tool called MEEGA +, it was pointed out that the game had an easy-to-learn interface, but that its challenges were very easy and did not create a good experience for the player. More refinement is needed for the adaptive model and the game design.

Cette thèse propose une méthode originale pour la génération automatique d'un code de simulation pour les systèmes à événements discrets. Cette méthode utilise l'information de localisation des produits lors du fonctionnement du système. Ce flux composé par des tuples (product id, location, time) constitue le point d'entrée pour l'algorithme proposé de génération d'un modèle de simulation de type réseau de files d'attente. Ce type d'approche permet, outre un gain important de temps pour la conception initiale du modèle, une maintenance et reconfiguration " on line " du modèle. La thèse est composée de 5 chapitres. Le premier chapitre pose la problématique et fixe le cadre théorique de la thèse. Le second chapitre est une revue de la littérature sur la simulation en général et sur les travaux utilisant la notion de trajectoires à des fins de modélisation. Le troisième chapitre sert à mettre en avant la proposition au coeur de cette thèse. Le quatrième chapitre décrit le générateur développé. Le cinquième et dernier chapitre présente les travaux d'expérimentation et de validation du générateur. / This thesis work proposes a novel method for the automatic generation of simulation code of discrete event systems. This method uses the location information of the products during operation of the system. This flow, of tuples (product id, location, time) is the input of the proposed algorithm that generates a queueing-network simulation model. This type of approach can achieve a significant time gain for the design, maintenance and on-line reconfiguration of the model. The thesis is composed on five chapters. The first chapter poses the problem and fixes the theoretical framework of the thesis. The second chapter is a literature review of the broad sense of simulation and present works using the notion of trajectories for modeling. The third chapter is to highlight the proposal of the heart of this thesis. The fourth chapter describes the generator developed. The fifth and final chapter presents the experiments and the validation of the generator

L'utilisation des véhicules sous-marins (ROV, AUV, gliders) s'est considérablement accrue ces dernières décennies, aussi bien dans le domaine de l'offshore ou de l'océanographie, que pour des applications militaires. Dans cette thèse, nous abordons le problème particulier de la commande des véhicules sous-marins à faible inertie et fort rapport puissance/inertie. Ces derniers constituent des systèmes fortement non linéaires, dont la dynamique est susceptible de varier au cours du temps (charge embarquée, caractéristiques des propulseurs, variation de salinité...) et qui sont très sensibles aux perturbations environnementales (chocs, traction sur l'ombilical...). Afin d'assurer des performances de suivi de trajectoire satisfaisantes, il est nécessaire d'avoir recours à une commande adaptative qui compense les incertitudes ou les variations des paramètres du modèle dynamique, mais également qui rejette les perturbations, telles que les chocs. A cette fin, nous proposons dans ce manuscrit, l'étude théorique et la validation expérimentale de plusieurs lois de commande pour véhicules sous-marins. Nous analysons tout d'abord des approches classiques dans ce domaine (commande PID et commande par retour d'état non linéaire), puis nous les comparons avec deux autres architectures de commande. La première est la commande adaptative L1 non linéaire, introduite en 2010 notamment pour la commande des véhicules aériens, et implémentée pour la première fois sur un véhicule sous-marin. Le découplage entre adaptation et robustesse permet l'utilisation de très grands gains d'adaptation (et donc une convergence plus rapide des paramètres estimés, sans aucune connaissance a priori), sans pour autant dégrader la stabilité. La seconde méthode, que nous proposons et qui constitue l'apport principal de cette thèse, est une évolution de la commande L1, permettant d'en améliorer les performances lors du suivi d'une trajectoire variable. Nous présentons une analyse de stabilité de cette commande, ainsi que sa comparaison expérimentale avec les autres lois de commande (commande PID, commande adaptative par retour d'état non linéaire et commande adaptative L1 standard). Ces expérimentations ont été réalisées sur un mini-ROV et plusieurs scenarii ont été étudiés, permettant ainsi d'évaluer, pour chaque loi, sa robustesse et son aptitude à rejeter les perturbations. / Underwater vehicles have gained an increased interest in the last decades given the multiple tasks they can accomplish in various fields, ranging from scientific to industrial and military applications. In this thesis, we are particularly interested in the category of vehicles having a high power to weight ratio. Different challenges in autonomous control of such highly unstable systems arise from the inherent nonlinearities and the time varyingbehavior of their dynamics. These challenges can be increased by the low inertia of this class of vehicles combined with their powerful actuation. A self tuning controller is therefore required in order to avoid any performance degradation during a specific mission. The closed-loop system is expected to compensate for different kinds of disturbances or changes in the model parameters. To solve this problem, we propose in this work the design,analysis and experimental validation of different control schemes on an underwater vehicle. Classical methods are initially proposed, namely the PID controller and the nonlinear adaptive state feedback (NASF) one, followed by two more advanced schemes based on the recently developed L1 adaptive controller. This last method stands out among the other developed ones in its particular architecture where robustness and adaptation are decoupled. In this thesis, the original L1 adaptive controller has been designed and successfullyvalidated then an extended version of it is proposed in order to deal with the observed time lags occurring in presence of a varying reference trajectory. The stability of this latter controller is then analysed and real-time experimental results for different operating conditions are presented and discussed for each proposed controller, assessing their performance and robustness.

A necessidade de jovens pesquisadores lerem correta e rapidamente uma grande quantidade de textos escritos em inglês, que é a língua franca da ciência, representa uma barreira considerável para eles. Dada essa necessidade, em 2001, o programa de mestrado em Ciências da Computação e Matemática Computacional do ICMC-USP passou a avaliar a proficiência em inglês dos alunos quanto à habilidade de reconhecerem o gênero de textos científicos em inglês, com as convenções que lhe são características. O site do Exame de Proficiência em Inglês (EPI)disponibiliza Exames Modelos com correção automatizada para os alunos saberem antecipadamente como será o exame formal, também informatizado. Porém, a prática com o Exame Modelo fornece apenas um relatório com o escore do aluno, não oferecendo meios para ele relembrar seus conhecimentos, receber instrução ou ainda rever conceitos errôneos. Uma possível solução para essa lacuna é um ambiente computacional que propicie um contexto favorável à aprendizagem do inglês instrumental. Na perspectiva de favorecer tal aprendizado e auxiliar os alunos a estudarem para o EPI é que essa pesquisa propôs um Ambiente Computacional de Aprendizagem (ACA) para o inglês instrumental. O ACA desenvolvido, denominado Computer-Aided Learning of English for Academic Purposes (CALEAP-Web), é fruto da integração de um Teste Adaptativo Informatizado (TAI), denominado de Adaptive English Proficiency Test for the Web (ADEPT), e um Ambiente Computacional de Tarefas (ACT), denominado de Computer-Aided Task Environment for Scientific English (CATESE). Um TAI consiste em itens (questões) selecionados de acordo com o nível de habilidade estimado do aluno, gerando um teste individualizado. Já o ACT fornece tarefas pré-determinadas para o aprendizado do aluno. Para o ADEPT foram realizados dois experimentos com alunos de mestrado em Ciências da Computação e Matemática Computacional do ICMC-USP e com o CALEAP-Web, como um todo, foram realizadas simulações para a verificação de sua operacionalidade. Embora o ACA tenha sido desenvolvido para apoiar os alunos do programa de mestrado do ICMC-USP, ele é suficientemente genérico para ser usado em outros programas de mestrado que avaliam o inglês instrumental e por jovens pesquisadores que desejam conhecer as convenções do gênero de textos científicos em inglês. / The need to read correctly and rapidly a huge amount of scientific papers, that is predominantly in English as the lingua franca for science, represents a considerable barrier for novice non-native English researchers. Due to this need, in 2001, the MSc. program in Computer Science and Computational Maths at ICMC-USP proposed a new type of proficiency exam to evaluate students\' competence in terms of the demands of highly standardized research articles written in English. The English proficiency exam website makes available simulated tests for the students to be aware of the contents and scoring procedure and to practice before applying for the formal exam, which is also automatically corrected. However, practice with simulated tests only provides a report with the student\'s score. There is no support to correct student\'s misconceptions, to instruct or to refresh prerequisite knowledge. A possible solution to fulfil this gap is to provide a computer-based environment designed to arouse students\' consciousness about the conventions in English usage in research papers. The purpose of this research was to propose such an environment. The computer-based environment developed, called Computer-Aided Learning of English for Academic Purposes (CALEAP-Web), is the result of the integration of an adaptive test, named Adaptive English Proficiency Test for the Web (ADEPT), and a task based environment called Computer-Aided Task Environment for Scientific English (CATESE). An adaptive test consists of items (questions) selected according to the estimated ability level of the student, providing an individualized test. A task based environment provides predefined tasks to ensure student\'s learning in a certain issue. Two evaluation experiments with students from the MSc. program in Computer Science and Computational Maths at ICMC-USP were carried out with ADEPT; with CALEAP-Web simulations were designed to verify functionality and robustness. Although CALEAP-Web was developed to support students from the MSc. program in Computer Science and Computational Maths at ICMC-USP, it is sufficiently generic to be used in other master programs which assess instrumental English and by novice researchers who need to know the demands of highly standardized research articles written in English.

Até 2020 espera-se que as redes celulares aumentam em dez vezes a área de cobertura, suporte cem vezes mais equipamentos de usuários e eleve a capacidade da taxa de dados em mil vezes, comparada as redes celulares atuais. A densa implantação de pequenas células é considerada uma solução promissora para alcançar essas melhorias, uma vez que aproximar as antenas dos usuários proporciona maiores taxas de dados, devido à qualidade do sinal em curtas distâncias. No entanto, operar um grande número de antenas pode aumentar significativamente o consumo de energia da infraestrutura de rede. Além disso, a grande inserção de novos rádios pode ocasionar maior interferência espectral entre as células. Nesse cenário, a gestão dos recursos de rádio é essencial devido ao impacto na qualidade do serviço prestado aos usuários. Por exemplo, baixas potências de transmissão podem deixar usuários sem conexão, enquanto altas potências elevam a possibilidade de ocorrência de interferência. Além disso, a reutilização não planejada dos recursos de rádio causa a ocorrência de interferência, resultando em baixa capacidade de transmissão, enquanto a subutilização de recursos limita a capacidade total de transmissão de dados. Uma solução para controlar a potência de transmissão, atribuir os recursos de rádio e garantir o serviço aos usuários é essencial. Nesta dissertação, é proposto um algoritmo adaptativo de Monte Carlo para realizar alocação global de recursos de forma eficiente em termos de energia, para arquiteturas Heterogeneous Cloud Radio Access Network (H-CRAN), projetadas como futuras redes de quinta geração (5G). Uma solução eficiente para a alocação de recursos em cenários de alta e baixa densidade é proposta. Nossas contribuições são triplas: (i) proposta de uma abordagem global para o problema de atribuição de recursos de rádio na arquitetura HCRAN, cujo caráter estocástico garante uma amostragem geral de espaço de solução; (ii) uma comparação crítica entre nossa solução global e um modelo local; (iii) a demonstração de que, para cenários de alta densidade, a Eficiência Energética não é uma medida adequada para alocação eficiente, considerando a capacidade de transmissão, justiça e total de usuários atendidos. Além disso, a proposta é comparada em relação a três algoritmos de alocação de recursos de última geração para redes 5G. / Up until 2020 it is expected that cellular networks must raise the coverage area in 10-fold, support a 100-fold more user equipments, and increase the data rate capacity by a 1000-fold in comparison with current cellular networks. The dense deployment of small cells is considered a promising solution to reach such aggressive improvements, once it moves the antennas closer to the users, achieving higher data rates due to the signal quality at short distances. However, operating a massive number of antennas can significantly increase the energy consumption of the network infrastructure. Furthermore, the large insertion of new radios brings greater spectral interference between the cells. In this scenery, the optimal management of radio resources turn an exaction due to the impact on the quality of service provided to the users. For example, low transmission powers can leave users without connection, while high transmission powers can contribute to inter radios interference. Furthermore, the interference can be raised on the unplanned reuse of the radio resources, resulting in low data transmission per radio resource, as the under-reuse of radio resources limits the overall data transmission capacity. A solution to control the transmission power, assign the spectral radio resources, and ensure the service to the users is essential. In this thesis, we propose an Adaptive Monte Carlo algorithm to perform global energy efficient resource allocation for Heterogeneous Cloud Radio Access Network (HCRAN) architectures, which are forecast as future fifth-generation (5G) networks. We argue that our global proposal offers an efficient solution to the resource allocation for both high and low density scenarios. Our contributions are threefold: (i) the proposal of a global approach to the radio resource assignment problem in H-CRAN architecture, whose stochastic character ensures an overall solution space sampling; (ii) a critical comparison between our global solution and a local model; (iii) the demonstration that, for high density scenarios, Energy Efficiency is not a well suited metric for efficient allocation, considering data rate capacity, fairness, and served users. Moreover, we compare our proposal against three state-of-the-art resource allocation algorithms for 5G networks.

La miniaturisation des technologies de semi-conducteurs a permis en quelques décennies de concevoir des systèmes toujours plus complexes, comprenant aujourd'hui plusieurs milliards de transistors sur un même substrat de silicium. Cette augmentation des densités d'intégration fait face à une contrainte physique représentée par la quantité de puissance consommée par unité de surface. À cela s'ajoutent également des problèmes de fiabilité, en raison notamment des hot-spots, qui peuvent accélérer la dégradation des transistors et réduire en conséquence la durée de vie du composant. L'efficacité énergétique des circuits devient un enjeu majeur, aussi bien dans le domaine de l'embarqué que pour des applications de calcul haute performance. La prise en compte de ces contraintes nécessite la mise en place de solutions nouvelles, s'appuyant notamment sur des techniques d'auto-adaptation. Celles-ci reposent généralement sur un processus bouclé en trois phases: (i) le monitoring qui consiste à observer l'état du système, (ii) le diagnostic qui analyse les informations relevées pour optimiser le fonctionnement du système, et (iii) l'action qui règle les paramètres en conséquence. L'efficacité d'une méthode d'adaptation dépend non seulement de l'algorithme d'optimisation mais aussi de la précision de l'information observée en ligne. Le monitoring est généralement effectué à l'aide d'un ensemble de capteurs intégrés (analogiques ou numériques). Les méthodes industrielles actuelles consistent à placer un nombre de capteurs par ressource (monitoring statique). Cependant, ces méthodes sont généralement très coûteuses et nécessitent l'insertion d'un grand nombre d'unités pour avoir une information précise sur le comportement du système à une résolution spatiale et temporelle fine. Cette thèse propose une approche innovante qui intervient en amont; un ensemble de techniques issues du domaine de la fouille de données est mis en œuvre pour l'analyse de données extraites des différents niveaux d'abstractions à partir du flot de conception, ce afin de définir une solution optimale en terme de coût et de précision. Notre méthode permet de dégager de manière systématique l'information pertinente requise pour la mise en œuvre d'un monitoring efficace et dans un contexte où la consommation et la fiabilité apparaissent comme de fortes contraintes, cette thèse s'intéresse plus particulièrement à celui de la puissance et de la température sur puce. / Over the last decades, the miniaturization of semiconductor technologies has allowed to design complex systems, including today's several billions of transistors on a single die. As a consequence, the integration density has increased and the power consumption has become significant. This is compounded by the reliability issues represented by the presence of thermal hotspots that can accelerate the degradation of the transistors, and consequently reduce the chip lifetime. In order to face these challenges, new solutions are required, based in particular on the self-adaptive systems. These systems are mainly composed of a control loop with three processes: (i) the monitoring which is responsible for observing the state of the system, (ii) the diagnosis, which analyzes the information collected and make decisions to optimize the behavior of the system, and (iii) the action that adjusts the system parameters accordingly. However, effective adaptations depend critically on the monitoring process that should provide an accurate estimation about the system state in a cost-effective way. The monitoring is typically done by using integrated sensors (analog or digital). The industrial methods consist of placing one sensor per resource (static monitoring). However, these methods are usually too expensive, and require a large number of units to produce a precise information at a fine-grained resolution. This thesis proposes an innovative and ‘upstream' approach; a set of data mining techniques is used to analyze data extracted from various levels of abstractions from the design flow, in order to define the optimum monitoring in terms of cost and accuracy. Our method systematically identifies relevant information required for the implementation of effective monitoring. This thesis mainly focuses on the monitoring of the power and the temperature of the chip.

Cette thèse étudie les aspects de contrôle d'applications optique adaptative, une technologie utilisée pour améliorer la performance des systèmes optiques en réduisant l'effet des distorsions de front d'onde, à l'imagerie haute résolution angulaire. Le problème Adaptive Optics contrôle est présenté à travers une revue de la littérature. Par conséquent, la conception d'un contrôleur de rétroaction est adressée, d'un point de vue moderne de contrôle, au moyen de la méthode de contrôle Linéaire Quadratique Gaussienne. L'approche proposée met l'accent sur la capacité de la boucle d'optique adaptative de rejeter l'aberration atmosphérique. On dérive un système de représentation diagonale état-espace qui sépare nettement la dynamique de la plante (miroir déformable et le capteur de front d'onde) de la dynamique des perturbations (modèle atmosphérique). Cette représentation facilite la résolution numérique du problème. Une analyse de fréquence est effectuée pour vérifier les spécifications de performance et de robustesse de la multiple-input multiple-système de rétroaction de sortie. De plus, nous analysons les performances et la robustesse de LQG contrôle basé par rapport au témoin intégrante classique, au moyen de bout en bout des simulations et en considérant les différents niveaux de bruit du capteur de front d'onde. Durant le-ciel observations, l'énergie turbulente et la vitesse relative de chaque couche de l'atmosphère peut changer rapidement dégrader l'estimation de front d'onde. Pour cette raison, un algorithme de modèle numérique de conception garantissant une performance satisfaisante rejet de perturbations, même dans le cas de variables dans le temps caractéristique de la turbulence est dérivé. Expériences numériques en utilisant les CAOS du progiciel ont été menées pour démontrer la robustesse de chaque approche proposée. Compte tenu de la conviction auteur que l'avenir de l'optique adaptative repose également sur le développement d'un plus sophistiqués (par exemple, non linéaire) des modèles, une quantité importante de travail a été consacrée à l'étude de deux classes de méthodes de reconnaissance des formes répandues. À savoir Support Vector Machines et méthodes du noyau, dont la régression des capacités sont exploitées dans la solution du problème non linéaire suivi optimal. En ce qui concerne Support Vector Machines, grâce à la théorie du contrôle optimal singulier, les contraintes se relâchent permettant une résolution plus facile et plus rapide numérique du problème d'optimisation. Alors que, dans le cas totalement déterministe du contrôleur Support Vector résulte plus simple à synthétiser. En référence aux méthodes du noyau, une tentative originale de réunir leurs forces de régression avec le concept de contrôle adaptatif inverse est présentée. Le noyau récursif des moindres carrés algorithme est utilisé pour mettre en œuvre un contrôleur adaptatif inverse capable de forcer une dynamique non linéaire appropriés pour suivre une sortie désirée. Cette méthode très peut également être utilisé pour vérifier si une trajectoire donnée arbitraire est une sortie admissible pour le système non linéaire à l'étude. Un tel algorithme innovant pourrait être utilement appliquée dans les travaux futurs, le contrôle de Tip-Tilt miroirs. Finalement, une première esquisse du cadre théorique soutenant l'utilisation du contrôle adaptatif inverse pour la solution du problème de suivi général est donné. Après l'introduction de la formulation mathématique du problème de suivi et les définitions nécessaires mathématiques, des conditions suffisantes et nécessaires (cas linéaire) et des conditions suffisantes (cas non-linéaire) de l'existence de la solution sont dérivés. / His thesis investigates the control aspects of Adaptive Optics applications, a technology used to improve the performance of optical systems by reducing the effect of wavefront distortions, to high angular resolution imaging. The Adaptive Optics control problem is presented through a survey of the literature. Consequently, the design of a feedback controller is addressed, from a modern control point of view, by means of the Linear Quadratic Gaussian control methodology. The proposed approach emphasizes the ability of the adaptive optics loop to reject the atmospheric aberration. We derive a diagonal state-space system representation which clearly separates the dynamics of the plant (deformable mirror and wavefront sensor) from the disturbance dynamics (atmospheric model). This representation facilitates the numerical resolution of the problem. A frequency analysis is carried out to check the performance and robustness specifications of the multiple-input multiple-output feedback system. Moreover, we analyze the performance and the robustness of LQG-based control compared to classic integral control, by means of end-to-end simulations and by considering different levels of wavefront sensor noise. During on-sky observations, the turbulent energy and relative speed of each atmospheric layer can change rapidly degrading the wavefront estimate. For this reason, a numerical model design algorithm guaranteeing satisfactory disturbance rejection performance even in the case of time-varying turbulence's characteristic is derived. Numerical experiments using the Software Package CAOS have been conducted to demonstrate the robustness of every proposed approach. Given the author firm belief that the future of Adaptive Optics also relies on the development of more sophisticated (i.e., nonlinear) models, a substantial amount of work was dedicated to the study of two classes of widespread pattern recognition methods. Namely Support Vector Machines and Kernel Methods, whose regression capabilities are exploited in the solution of the nonlinear optimal tracking problem. Concerning Support Vector Machines, thanks to the singular optimal control theory, constraints are loosened permitting an easier and faster numerical resolution of the optimization problem. So that, in the fully deterministic case the Support Vector controller results simpler to synthesize. With reference to Kernel Methods, an original attempt to bring together their regression strengths with the concept of Adaptive Inverse Control is presented. The Kernel Recursive Least-Square algorithm is used to implement an adaptive inverse controller capable of forcing a suitable nonlinear dynamics to follow a desired output. This very method can also be used to check if a given arbitrary trajectory is an admissible output for the nonlinear system under study. Such an innovative algorithm could be fruitfully applied, in future works, to the control of Tip-Tilt mirrors. Eventually, a first sketch of the theoretical framework supporting the use of Adaptive Inverse Control for the solution of the general tracking problem is given. After introducing the mathematical formulation of the tracking problem and the needed mathematical definitions, sufficient and necessary conditions (linear case) and sufficient (nonlinear case) conditions to the existence of the solution are derived.

A busca por substâncias imunomoduladoras oriundas de plantas medicinais é crescente devido à grande variedade de doenças relacionadas às respostas imunológicas desordenadas, como as autoimunidades e o câncer. Neste sentido, sabe-se que a Paullinia cupana, popularmente conhecida como guaraná, é uma planta medicinal que apresenta entre outros efeitos, propriedades quimiopreventiva e antitumoral em camundongos. Entretanto, até o presente estudo, não havia relatos na literatura a respeito dos efeitos desta planta sobre o sistema imune, conhecido por desempenhar papel fundamental não só no controle de infecções, mas também no combate ao câncer. Assim, o presente estudo avaliou os efeitos do pó das sementes de guaraná e de seus principais compostos ativos, cafeína e catequina, sobre o sistema imune de camundongos C57BL/6 através de protocolos preconizados pela Organization for Economic Cooperation and Development (OECD) com modificações. Os resultados obtidos mostraram que o guaraná, devido à ação majoritária da catequina, mas com a participação da cafeína também, diminui as respostas imunes, celular e humoral, verificada através da diminuição da expansão de linfócitos T frente a um antígeno específico, redução da resposta de hipersensibilidade do tipo tardia e título de anticorpos, diminuição dos pesos relativos do baço e timo e celularidade deste último, além de aumento de células imaturas no timo. Desta forma, os resultados aqui expostos demonstram, pela primeira vez, que a ingestão de guaraná reduz a imunidade adaptativa através da diminuição das respostas celular e humoral. Assim, futuros estudos deverão ser realizados, no sentido de se verificar a possibilidade do uso desta planta ou de seus princípios ativos no controle de doenças inflamatórias e autoimunes, caracterizadas por uma resposta exacerbada do sistema imune. / Seeking for immunomodulatory compounds derived from medicinal plants is increasing due to the huge variety of diseases related to disordered immune responses, such as autoimmunity and cancer. Therefore, it is known that the Paullinia cupana, popularly known as guarana, is a medicinal plant that has among other effects, chemopreventive and antitumor properties in mice. However, until the present moment, there were no reports in the literature about the effects of this plant on the immune system, which is known to play key role not only in the infectious control, but also in combat cancer. Thus, this study evaluated the effects of guarana seed powder and its major active principles, caffeine and catechin, on the immune system of C57BL/6 mice through protocols recommended by the Organization for Economic Cooperation and Development (OECD) with modifications. The results showed that guarana, due to the majority action of catechin, but also with the participation of caffeine, decreases the cellular and humoral immune responses, as indicated by the decreased expansion of antigen-specific T lymphocyte, reduced response delayed-type hypersensitivity and titer of antibodies, reduction in relative weights of thymus and spleen cellularity, besides increase of immature cells in the thymus. Thus, the results showed here demonstrate, for the first time, that the intake of guarana decreases the adaptive immunity by reducing cellular and humoral responses. Thus, futures studies should be conducted in order to better evaluate the use of this plant and/or its main active principle in the treatment of inflammatory and autoimmune diseases, which are characterized by exarcebated immune response.