Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

La thèse traite du problème de la reconnaissance des gestes avec des accents particuliers orientés vers la modélisation des trajectoires de mouvement ainsi que vers l’estimation de la variabilité présente dans l’exécution gestuelle. Les gestes sont acquis dans un scénario typique pour la vision par ordinateur qui approche les particularités des surfaces interactives. On propose un modèle flexible pour les commandes gestuelles à partir d’une représentation par courbes splines et des analogies avec des éléments de la théorie d’élasticité de la physique classique. On utilise les propriétés du modèle pour la reconnaissance des gestes dans un contexte d’apprentissage supervisé. Pour adresser le problème de la variation présente dans l’exécution des gestes, on propose un modèle qui mesure dans une manière quantitative et objective les tendances locales que les utilisateurs sont tentés d'introduire dans leurs exécutions. On utilise ce modèle pour proposer une solution à un problème reconnu comme difficile dans la communauté : la segmentation automatique des trajectoires continues de mouvement et l’identification invariante a l’échelle des commands gestuelles. On démontre aussi l’efficacité du modèle pour effectuer des analyses de type ergonomique pour les dictionnaires de gestes. / We address in this thesis the problem of gesture recognition with specific focus on providing a flexible model for movement trajectories as well as for estimating the variation in execution that is inherently present when performing gestures. Gestures are captured in a computer vision scenario which approaches somewhat the specifics of interactive surfaces. We propose a flexible model for gesture commands based on a spline representation which is enhanced with elastic properties in a direct analogy with the theory of elasticity from classical physics. The model is further used for achieving gesture recognition in the context of supervised learning. In order to address the problem of variability in execution, we propose a model that measures objectively and quantitatively the local tendencies that users introduce in their executions. We make use of this model in order to address a problem that is considered hard by the community: automatic segmentation of continuous motion trajectories and scale invariant identification of gesture commands. We equally show the usefulness of our model for performing ergonomic analysis on gesture dictionaries.

A gestão da manutenção de uma edificação tem influência direta em seu desempenho, ou seja, em sua duração, seus impactos e seu consumo de recursos como água e energia. O planejamento, a facilidade e demanda de execução dos serviços de manutenção, assim como a transparência para os usuários das necessidades, prazos e procedimentos de manutenção são de fundamental importância para a qualidade e sustentabilidade dos edifícios. O processo de gestão da manutenção de uma edificação tem início ainda nas etapas iniciais de projeto do empreendimento, quando tomadas de decisões que considerem fatores de manutenção fornecem subsídios para elaboração de projetos da manutenção mais eficazes e de menor impacto social, ambiental e econômico, tem continuidade durante a etapa de construção quando se deve primar pela qualidade e continua durante a etapa de operação, uso e manutenção. Particularmente em empreendimentos de habitação de interesse social, a geração de subsídios, a consideração de fatores de manutenção, o planejamento da manutenção, da maneira como operacionalizá-la, são fatores determinantes para conservação de edificações, além de contribuir com a luta contra o déficit habitacional, buscando empreendimentos de melhor qualidade, com maiores graus de manutenibilidade e conseqüentemente tarifas condominiais e consumo de recursos reduzidos. Neste contexto, esta pesquisa tem como objetivo contribuir com o processo de gestão da manutenção, a partir de um levantamento feito através do ponto de vista das empresas administradoras de empreendimentos PAR do estado de São Paulo, abordando o tratamento dispensado à manutenção a partir da fase de uso e operação, identificando requisitos para elaboração de diretrizes que contribuam com a operacionalização da manutenção. / A building maintenance management has direct influence on its performance, ie, its duration, its impacts and its resources consumption as water and energy. The planning, execution ease and demand of maintenance services as well as transparency of needs, deadlines and maintenance procedures for the users are essential to the quality and sustainability of buildings. The process of a building maintenance management begins in the early stages of a building design, when making decisions considering maintenance aspects provides subsides for more effective designs with less social, environmental and economic impacts. This process continues during the construction stage when it is necessary to strive for quality, maintaining this line of reasoning during the operation, use and maintenance phase. Particularly in housing projects, generating subsides, considering maintenance factors, maintenance planning and the means to operate it are determining factors for building maintenance, thus it contributes to combating the housing shortage seeking buildings of better quality, with greater degrees of maintainability and condominium prices and therefore reduced resource consumption. Thus, this research aims to contribute to the maintenance management process, starting from a survey done by the point of view of companies that manage PAR housing projects of the state of São Paulo, addressing the treatment given to the operation, use and maintenance phase, identifying requirements for guidelines developing to contribute to the maintenance operation and management.

La polarité cellulaire est un processus fondamental qui contrôle les spécificités fonctionnelle et physiologique de la plupart des cellules eucaryotes. Cette asymétrie intracellulaire repose sur l’existence de compartiments membranaires distincts, à la fois dans leur composition en protéines mais également en phosphatidyl-inositols (PIs). Ainsi, la mise en place et le maintien de la localisation asymétrique de modules multi-protéiques associés notamment aux protéines PAR sont essentiels pour l’élaboration des domaines de polarité cellulaire. Durant ma thèse, j’ai étudié les relations entre les protéines de polarité et les PIs dans le contrôle de la polarité cellulaire. Plus particulièrement, en utilisant la chambre ovarienne de Drosophile, j’ai cherché à caractériser la suite d’évènements qui en amont régule l’activité de la PIP5K, Skittles (SKTL), qui produit le PI(4,5)P2 et à caractériser les mécanismes moléculaires qui lient le PI(4,5)P2, SKTL et les protéines PAR dans le contrôle et le maintien de la polarité cellulaire. J’ai contribué à caractériser l’importance de PI(4,5)P2 majoritairement produit par SKTL, dans le maintien de la polarité apico-basale et lors de la morphogenèse des cellules folliculaires de la chambre ovarienne. Le PI(4,5)P2 assure la localisation apicale de PAR3 et le maintien des jonctions adhérentes, sans affecter la localisation de PAR1. Par une méthode de quantification précise, j’ai ensuite démontré dans l’ovocyte que SKTL et le PI(4,5)P2, probablement grâce au trafic vésiculaire, étaient requis pour à la fois l’accumulation à l’antérieur de PAR3 et son exclusion au postérieur qui se fait à partir du stade 9B. L’accumulation antérieure de PAR3 est également dépendante d’un transport Dynéine dépendant et de la kinase IKKε tandis que son exclusion postérieure dépendant des phosphorylations par PAR1. Enfin, j’ai également étudié les modifications post traductionnelles de SKTL et leur importance dans la polarité cellulaire. J’ai identifié la présence de palmitoylation et de phosphorylations dont certaines impliquent la kinase PAR1 et la phosphatase PP1. Ces phosphorylations pourraient avoir un lien avec le rôle de SKTL dans le trafic vésiculaire. Ces résultats permettent donc d’élucider certains mécanismes cellulaires qui contrôlent la mise en place et le maintien de la polarité des cellules en liant les PIs et les protéines PAR / Cell polarity is a fundamental process that controls cell’s functional and physiological specificities. This process relies on membranous compartments differently composed both on proteins and on phosphatidyl-inositols (PIs). Indeed, through their asymmetric localization, polarity proteins, such as the PAR proteins, are essentials to establish and maintain polarity of the cells. During my PhD, I studied the interplay between the polarity proteins and the PIs. Using the Drosophila egg chamber, as a model, I aimed to characterized the upstream events that regulate the PI(4,5)P2 producing kinase (PIP5K), Skittles (SKTL), activity and localization. I also studied the downstream molecular process that link the PI(4,5)P2, SKTL and the PAR proteins in cell polarity. I contributed to the characterization of the importance of PI(4,5)P2, mainly produced by SKTL in maintaining the apical-basal polarity and during the morphogenesis of the follicle cells. The PI(4,5)P2 is ensuring PAR3 and adherens junctions but not PAR1 proper localizations. Next, through a precise quantification method, I showed that SKTL and the PI(4,5)P2, probably via vesicular traffic, were also ensuring PAR3 proper localizations (anterior accumulation and stage 9B posterior exclusion) in the oocyte. PAR3 accumulation also relies on a Dynein mediated transport and the IKKε kinase while its posterior exclusion relies on PAR1 phosphorylation. Finally, I studied SKTL post translational modifications and their relevance on cell polarity. I identified palmitoylation and phosphorylations that are regulated by the kinase PAR1 and the phosphatase PP1. SKTL phosphorylations seem to be related to its role on the vesicular traffic. Altogether these results clarify some mechanisms involving both PIs and PAR proteins in cell polarity maintaining and establishment

Pesquisas recentes investigando o receptor ativado por protease do tipo 2 (PAR-2) sugerem uma associação entre este receptor e a inflamação periodontal. Além disso, é sabido que a gingipaína, protease bacteriana secretada por um importante periodontopatógeno, Porphyromonas gingivalis (Pg), tem a capacidade de ativar o PAR-2. Ademais, um estudo anterior do grupo, verificou que quanto mais profunda a bolsa periodontal, maior era a expressão do receptor PAR-2. No entanto, não é sabido se a expressão de PAR-2 é proporcional a severidade de doença periodontal e a quantidade de gingipaína expressa na bolsa periodontal. Desta forma, o presente estudo, verificou no fluido gengival a correlação entre a expressão gênica de PAR-2 (Real Time-PCR) com os parâmetros clínicos periodontais, e a expressão gênica da protease gingipaína em pacientes com periodontite crônica severa e moderada, antes e após o tratamento periodontal não-cirúrgico. A expressão de PAR-2 e da protease gingipaína foi estatisticamente maior nos pacientes do grupo periodontite crônica severa (PS) em comparação com os pacientes do grupo periodontite crônica modera (PM) e controle (C). Além disso, o tratamento periodontal levou à redução significativa (p<0.05) da expressão de PAR-2 nos pacientes com periodontite crônica moderada. Em conclusão, dentro dos limites do presente estudo, nós demonstramos que a severidade da doença periodontal e a expressão de gingipaína influenciaram a expressão de PAR-2 no fluido gengival de pacientes com periodontite crônica. / Recent studies investigating the protease-activated receptor type 2 (PAR-2) suggest an association between the receptor and periodontal inflammation. In addition, it is known that gingipain, a bacterial protease secreted by an important periodontopathogen, Porphyromonas gingivalis (Pg), has the ability to activate PAR- 2. Furthermore, a previous study from our group found that the deeper the periodontal pocket, the higher the expression of the PAR-2 receptor. However, it is not known whether the expression of PAR-2 is associated to the severity of periodontal disease and the amount of gingipain expressed in the periodontal pocket. Thus, the present study verified, in the gingival fluid, the correlation between the PAR-2 gene expression (Real Time-PCR) with the clinical periodontal parameters, and the gene expression of the gingipain protease in patients with moderate and severe chronic periodontitis before and after non-surgical periodontal treatment. PAR-2 expression and gingipain protease were statistically more expressed in patients of the severe chronic periodontitis group (PS) compared with those in the moderate chronic periodontitis group (PM) and control group (C). Furthermore, periodontal treatment led to a significant reduction (p <0.05) in the expression of PAR-2 in patients with moderate chronic periodontitis. In conclusion, within the limits of the present study, we demonstrated that the severity of periodontal disease and the expression of gingipain influenced the PAR-2 expression in the gingival fluid of patients with chronic periodontitis.

O presente trabalho tem por finalidade o uso da abordagem bayesiana para a estimativa de parâmetros sazonais dos modelos periódicos auto-regressivos (PAR). Após a determinação dos estimadores bayesianos, estes são comparados com os estimadores de máxima verossimilhança. A previsão para 12 meses é realizada usando os dois estimadores e os resultados comparados por meio de gráficos, tabelas e pelos erros de previsão. Para ilustrar o problema as séries escolhidas foram as séries hidrológicas da Usinas Hidroelétricas de Furnas e Emborcação. Tais séries foram selecionadas tendo em vista a necessidade de previsões com reduzido erro já que o sistema de operação das usinas hidroelétricas depende muito da quantidade de água existente em seus reservatórios e de planejamento e gerenciamento eficazes. / The objective of this research is to use bayesian method to estimate of sazonal parameters of periodic autoregressive models (PAR). The bayesian estimators are then compared with maximum likelihood estimators. The forecast for 12 months is made by using two estimators and comparing their results though graphs, tables and forecast error. The hydrological time series chosen were from Furnas and Emborcação Hydroeletric Power Plant. These series were chosen having in mind the necessity of series with reduced error in their forecast because system of operation in the Hydroeletric Power Plant depends on the quantity of the water in their resevoirs, eficient planning and management.

O presente trabalho tem por finalidade o uso da abordagem bayesiana para a estimativa de parâmetros sazonais dos modelos periódicos auto-regressivos (PAR). Após a determinação dos estimadores bayesianos, estes são comparados com os estimadores de máxima verossimilhança. A previsão para 12 meses é realizada usando os dois estimadores e os resultados comparados por meio de gráficos, tabelas e pelos erros de previsão. Para ilustrar o problema as séries escolhidas foram as séries hidrológicas da Usinas Hidroelétricas de Furnas e Emborcação. Tais séries foram selecionadas tendo em vista a necessidade de previsões com reduzido erro já que o sistema de operação das usinas hidroelétricas depende muito da quantidade de água existente em seus reservatórios e de planejamento e gerenciamento eficazes. / The objective of this research is to use bayesian method to estimate of sazonal parameters of periodic autoregressive models (PAR). The bayesian estimators are then compared with maximum likelihood estimators. The forecast for 12 months is made by using two estimators and comparing their results though graphs, tables and forecast error. The hydrological time series chosen were from Furnas and Emborcação Hydroeletric Power Plant. These series were chosen having in mind the necessity of series with reduced error in their forecast because system of operation in the Hydroeletric Power Plant depends on the quantity of the water in their resevoirs, eficient planning and management.

Pesquisas recentes investigando o receptor ativado por protease do tipo 2 (PAR-2) sugerem uma associação entre este receptor e a inflamação periodontal. Além disso, é sabido que a gingipaína, protease bacteriana secretada por um importante periodontopatógeno, Porphyromonas gingivalis (Pg), tem a capacidade de ativar o PAR-2. Ademais, um estudo anterior do grupo, verificou que quanto mais profunda a bolsa periodontal, maior era a expressão do receptor PAR-2. No entanto, não é sabido se a expressão de PAR-2 é proporcional a severidade de doença periodontal e a quantidade de gingipaína expressa na bolsa periodontal. Desta forma, o presente estudo, verificou no fluido gengival a correlação entre a expressão gênica de PAR-2 (Real Time-PCR) com os parâmetros clínicos periodontais, e a expressão gênica da protease gingipaína em pacientes com periodontite crônica severa e moderada, antes e após o tratamento periodontal não-cirúrgico. A expressão de PAR-2 e da protease gingipaína foi estatisticamente maior nos pacientes do grupo periodontite crônica severa (PS) em comparação com os pacientes do grupo periodontite crônica modera (PM) e controle (C). Além disso, o tratamento periodontal levou à redução significativa (p<0.05) da expressão de PAR-2 nos pacientes com periodontite crônica moderada. Em conclusão, dentro dos limites do presente estudo, nós demonstramos que a severidade da doença periodontal e a expressão de gingipaína influenciaram a expressão de PAR-2 no fluido gengival de pacientes com periodontite crônica. / Recent studies investigating the protease-activated receptor type 2 (PAR-2) suggest an association between the receptor and periodontal inflammation. In addition, it is known that gingipain, a bacterial protease secreted by an important periodontopathogen, Porphyromonas gingivalis (Pg), has the ability to activate PAR- 2. Furthermore, a previous study from our group found that the deeper the periodontal pocket, the higher the expression of the PAR-2 receptor. However, it is not known whether the expression of PAR-2 is associated to the severity of periodontal disease and the amount of gingipain expressed in the periodontal pocket. Thus, the present study verified, in the gingival fluid, the correlation between the PAR-2 gene expression (Real Time-PCR) with the clinical periodontal parameters, and the gene expression of the gingipain protease in patients with moderate and severe chronic periodontitis before and after non-surgical periodontal treatment. PAR-2 expression and gingipain protease were statistically more expressed in patients of the severe chronic periodontitis group (PS) compared with those in the moderate chronic periodontitis group (PM) and control group (C). Furthermore, periodontal treatment led to a significant reduction (p <0.05) in the expression of PAR-2 in patients with moderate chronic periodontitis. In conclusion, within the limits of the present study, we demonstrated that the severity of periodontal disease and the expression of gingipain influenced the PAR-2 expression in the gingival fluid of patients with chronic periodontitis.

Le groupe de Thompson projectif $T$ est l'ensemble des homéomorphismes du bord du disque hyperbolique qui sont $PSL_2((\bf Z))$ par morceaux avec points de rupture rationnels. Pour un sous-groupe $\Gamma$ de $PSL_2((\bf Z))$ on peut construire le sous-groupe $T_(\Gamma)$ de $T$ des homéomorphismes $\Gamma$ par morceaux, et on se demande si la propriété fondamentale de $T$ d'être de type fini est conservée. Cette étude dépend du genre de la surface associée à $\Gamma$. Le but principal de notre travail est de prouver qu'en genre nul, $T_(\Gamma)$ est de présentation finie (Peter Greenberg a montré qu'en genre strictement positif $T_(\Gamma)$ n'est pas de type fini). Nous commençons par conjuguer $T_(\Gamma)$ à un groupe d'homéomorphismes affines par morceaux dont nous prouvons, à l'aide de groupes de Thompson classiques, qu'il est de type fini. Puis nous donnons une description combinatoire de $T_(\Gamma)$ par des couples de forêts infinies, description qui nous permet de déterminer une présentation infinie régulière du groupe, puis une présentation finie.

The foundation of term rewriting is equational logic but for the sake of efficiency, the equations are oriented and become the rules of a term rewriting system. Term rewriting form a model of computation on algebraic data structures (terms). Term rewriting systems play an important role in various domains of computer science such as automated theorem proving, functional programming, code generation, problem formalization (security of cryptographic protocols). Rewriting starts with a ground term, and consists of repeatedly replacing a redex (an instance of a left-hand side) by its contractum (the corresponding right-handside after applying the substitution). Rewriting may eventually yield a term in normal form which is a term containing no redex. Natural questions in term rewriting are: * is the system terminating" (i.e. there are no infinite rewrite sequences)? * "is the system confluent" (if a term rewrites independently to two terms t1 and t2 , there exists a term s such that both t1 and t2 rewrite to s)? We are interested in systems which can be used as programs so we want to allow non-terminating computations. Confluence implies unicity of normal forms but does not imply termination. Confluent systems form a good framework for deterministic programming. They have the power of Turing machines. However confluence is not a decidable property for term rewriting systems. Orthogonal systems (i.e. linear and non-overlapping left-hand sides) which are always confluent form the framework of all this work, although some results may apply to the more general class of left-linear systems (linear left-hand sides). The first point we want to address is "how to compute the normal form?" and not end up in an infinite computation when the normal form exists. The second is "how to do that efficiently?". The following theorem of Huet and Levy [HL91] forms the basis of all result on optimal normalizing rewrite strategies for orthogonal term rewrite systems: "Every reducible term contains a needed redex, i.e., a redex which is contracted in every rewrite sequence to normal form, and repeated contraction of needed redexes results in a normal form, if the term under consideration has a normal form". Unfortunately, needed redexes are not computable in general. Hence, in order to obtain a computable optimal rewrite strategy, we are left to find (1) decidable approximations of neededness and (2) decidable properties of rewrite systems which ensure that every reducible term has a needed redex identi-fied by (1). Starting with the seminal work of Huet and Levy [HL91] on strong sequentiality, these issues have been extensively investigated in the literature [Com00, Jac96b, JS95, KM91, NST95, Oya93, Toy92]. In all these works Huet and Levy's notions of index, omega-reduction, and sequentiality figure prominently. We present here our contributions to this domain.