O crescente aumento no volume de dados complexos tem se tornado um desafio para pesquisadores. Séries temporais são um tipo de dados complexos que tem tido um crescimento em sua relevância, devido a sua importância para o monitoramento e acompanhamento de safras agrícolas. Assim, a mineração de informação a partir de grandes volumes de séries temporais para o apoio a tomada de decisões tem se tornado uma atividade valiosa. Uma das atividades importantes na mineração em séries temporais é a descoberta de padrões frequentes. Entretanto, a complexidade dessa atividade requer métodos rápidos e eficientes. Nesse contexto, esta dissertação de mestrado apresenta propostas para novos algoritmos e métodos para minerar e indexar séries temporais. Uma das propostas dessa dissertação é o índice Telesto, que utiliza uma estrutura baseada em árvores de sufixo generalizada para recuperar séries temporais em uma base de dados de séries temporais de modo rápido e eficiente. Outra proposta dessa dissertação é o algoritmo TrieMotif, que se baseia em uma trie para eliminar comparações desnecessárias entre subsequências, agilizando o processo de mineração de padrões frequentes em séries temporais. Os algoritmos propostos foram utilizados para a análise de dados climáticos e agrometeorológicos. Os resultados apresentados nessa dissertação de mestrado mostram que os algoritmos são escaláveis, podendo ser utilizados para grandes volumes de dados / Dealing with large volumes of complex data is a challenging task that has motivated many researchers around the world. Time series is a type of complex data that is growing in importance due to the increasing demand of sensors for surveillance and monitoring. Thus, mining information from large volumes of time series to support decision making is a valuable activity nowadays. This Master dissertation goes in this direction, as it proposes new algorithms and methods to mine and index time series. The novelty of the TrieMotif, a new algorithm to mine frequent patterns (motifs) from time series employing a trie structure that allows clever comparison between the sequences, as well as the Telesto index structure based on suffix trees area presented and discussed in the context of agrometeorological and climatological data, being the two main contributions of this work. The dissertation shows that the proposed algorithms are scalable, being suitable to big data, and when compared to the competitors they always presented the best results

Um dos grandes desafios em mineração de dados é a integração de dados temporais ao seu processo. Existe um grande número de aplicações emergentes que envolvem dados temporais, incluindo a identificação de transações fraudulentas em cartões de crédito e ligações telefônicas, a detecção de intrusão em sistemas computacionais, a predição de estruturas secundárias de proteínas, a análise de dados provenientes de sensores, entre muitas outras. Neste trabalho, tem-se interesse na classificação de séries temporais que representam sinais de áudio. Como aplicação principal, tem-se interesse em classificar sinais de insetos coletados por um sensor óptico, que deve ser capaz de contar e classificar os insetos de maneira automática. Apesar de serem coletados opticamente, os sinais capturados se assemelham a sinais de áudio. O objetivo desta pesquisa é comparar métodos de classificação por similaridade e por extração de atributos que possam ser utilizados no contexto da classificação de insetos. Para isso, foram empregados os principais métodos de classificação de sinais de áudio, que têm sido propostos para problemas como reconhecimento de instrumentos musicais, fala e espécies animais. Neste trabalho, é mostrado que, de modo geral, a abordagem por extração de atributos é mais eficaz do que a classificação por similaridade. Mais especificamente, os melhores resultados são obtidos com a utilização de coeficientes mel-cepstrais. Este trabalho apresenta contribuições significativas em outras aplicações, também relacionadas à análise de séries temporais e sinais de áudio, por similaridade e por extração de atributos / One of the major challenges in data mining is the integration of temporal data to its process. There are a number of emerging applications that involve temporal data, including fraud detection in credit card transactions and phone calls, intrusion detection in computer systems, the prediction of secondary structures of proteins, the analysis of data from sensors, and many others. In this work, our main interest is the classification of time series that represent audio signals. Our main interest is an application for classifying signals of insects collected from an optical sensor, which should count and classify insects automatically. Although these signals are optically collected, they resemble audio signals. The objective of this research is to compare classification methods based on similarity and feature extraction in the context of insects classification. For this purpose, we used the main classification methods for audio signals, which have been proposed for problems such as musical instrument, speech and animal species recognition. This work shows that, in general, the approach based on feature extraction is more accurate than the classification by similarity. More specifically, the best results are obtained with mel-frequency cepstrum coefficients. This work also presents significant contributions in other applications, also related to the analysis of time series and audio signals by similarity and feature extraction

Orientador: Tatiana Noronha de Souza / Resumo: O ensino de História no Brasil pode ser caracterizado por suas tradições inventadas, conceito definido por Hobsbawm. Em seu processo de constituição como código disciplinar, a História esteve vinculada à ideia de nacionalidade, cuja função foi normatizar e legitimar a estrutura social brasileira desigual. Foi desde o século XIX uma composição elitista - uma construção intelectual baseada em figuras selecionadas como os fundadores da nacionalidade, articulados dentro de uma totalidade hermética e inevitável cujo ente principal era a figura do Estado centralizado e sua função a manutenção do status quo. Contudo, desde o final do regime de exceção ocorrido nos anos 1980, o ensino de História passa por um processo de reconstituição e desenvolve entre outras questões a crítica a esta ideia de um passado homogêneo e harmonioso. Como não se trata de uma situação resolvida, ainda é possível a existência de reminiscências desde os primeiros anos de escolarização por conta das condições educacionais brasileiras. Nesta linha, esta pesquisa busca investigar os resultados de pesquisas sobre a atividade do professor que ensina História nos primeiros anos do Ensino Fundamental em teses e dissertações que realizaram pesquisa de campo, tentando definir as principais dificuldades deste profissional, que, em geral, não possui formação específica e verificar se ocorre a manutenção de temáticas que exaltam a mitologia da pátria, visto que funcione como mecanismo de harmonização social. A discussã... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: History teaching in Brazil can be characterized for its invented traditions, concept defined by Hobsbawm. In its constitution process as a school subject, History had been linked to the idea of nationality, which function was normalizing and legitimizing the unequal Brazilian social structure. It was since XIX Century an elitist composition – an intellectual construction based on chosen figures such as the founders of nationality, articulated within a hermetic and inevitable totality which principal entity was the centralized State and its function: the sustenance of the status quo. However, since the end of the exception regime occurred in the 1980s, History teaching passes by a process of reconstitution and development, among other questions, the critic to this idea of a homogeneous and harmonious past. As this is not a resolved situation, it’s still possible the existence of reminiscences since the first years of schooling because of the Brazilian educational conditions. In this line, this research seeks to investigate the results of researches about the activity of the professional who teaches History in the early grades of the elementary school in theses and dissertations that conducted field research, trying to define the principal difficulties of this professional and verifying if occurred the maintaining of subjects that exalt the homeland mythology. The discussion has its beginning in the historical presentation about how the History teaching was structured in Brazil... (Complete abstract click electronic access below) / Mestre

Les prétentions esthétiques de la série télévisée Les Soprano nous ont convaincu qu’elle s’inscrit dans une période significative de l’histoire des médias : la migration du cinéma et la de littérature vers la sérialité télévisée, avec tout ce que cela implique de transformations sociale, esthétiques et économiques fondamentales. Nous mettrons cette période en parallèle avec celle du cinéma du vingtième siècle et du roman du dix-neuvième siècle, afin de montrer que l’œuvre est habitée, de manière aussi bien volontaire qu’involontaire, par certaines contradictions multidimensionnelles inhérentes aux mutations de l’industrie culturelle. Ces contradictions impliquent la contamination pathologique du récit par l’interruption et la répétition, connotant une passivité fondamentale de l’Homme moderne. / The aesthetic pretensions of the television series The Sopranos have convinced us that it is part of a significant period in the history of the media: the migration of cinema and literature towards televised seriality with all that this implies of social, aesthetic and economic transformations. We will put this period in parallel with that of the twentieth century cinema and the nineteenth century novel to show that the work is inhabited, both voluntarily and involuntarily, by certain multidimensional contradictions inherent in the mutations cultural industry. These contradictions imply the pathological contamination of the story by interruption and repetition, which connotes a fundamental passivity of modern man.

Le Cameroun est un pays engagé dans une stratégie de développement volontariste, faisant toutefois face à d’importantes difficultés de croissance économique et à une corruption chronique. Suivant ce contexte, la présente étude a pour objet d’effectuer une analyse théorique et empirique des effets de la corruption sur la croissance économique du Cameroun. A cet effet, notre premier axe d’analyse vise à déterminer la nature de la corrélation entre la corruption et la croissance économique au Cameroun. Notre étude révèle qu’à long terme, la corruption affecte négativement le taux de croissance économique, suivant des effets directs d’une part, et suivant des effets indirects d’autre part, à travers l’investissement privé, le niveau d’éducation et les dépenses publiques. Dès lors, un deuxième axe d’analyse nous a paru intéressant à aborder, cherchant cette fois à évaluer l’effet indirect de la corruption sur la croissance économique via les distorsions dans la répartition des dépenses publiques. Les résultats obtenus ici montrent que la corruption provoque des distorsions tendant d’une part, à accroître les allocations aux des dépenses publiques d’investissement, pour un effet induit non significatif sur la croissance économique, et d’autre part, à réduire la part des dépenses publiques de fonctionnement, pour un effet induit négatif sur la croissance économique. / Cameroon is an economy committed to an expansionist development strategy, yet faced with major economic growth challenges and chronic corruption. In this context. The purpose of this study is, therefore, to carry out a theoretical and empirical analysis of the effects of corruption on Cameroon's economic growth. The first analysis framework is aimed at determining the nature of causal relationship between corruption and economic growth in Cameroon. The estimations reveal that in long run, an increase in the level of corruption leads to a reduction in the rate of economic growth. This negative relationship is following a direct effect as well as indirect effects through private investment, education and public spending. The second axis of analysis seeks to assess the indirect effect of corruption on economic growth through distortions in the distribution of public expenditure. Our results show that corruption brings about distortions aimed at increasing allocations to public investment spending thereby producing an insignificant effect on economic growth on the one hand, and reducing the share of public recurrent expenditure with a negative effect on economic growth, on the other hand.

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

Cette thèse propose une collection des nouveaux outils pour la représentation musicale. Ces modèles ont deux caractéristiques principales. D'un côté, ils sont inspirés par la géométrie et la topologie. De l'autre côté, ils ont une basse dimensionnalité, afin de garantir une visualisation intuitive des caractéristiques musicales qu'ils représentent. On s'est attaqué au problème de l'analyse musicale à partir de trois points de vue. On a représenté le contrepoint en utilisant des séries temporelles multivariées de matrices de permutations partielles. On a visualisé la conduite des voix en utilisant une classe particulière des tresses partielles et singulières. On donne ensuite une interpretation du Tonnetz comme complex simplicial et on utilise l'homologie persistante, afin de classifier des formes obtenues en déformant les sommets du Tonnetz. Ces déformations sont induites soit par des fonctions qui prennent en compte la nature symbolique de la musique, soit l'interaction symbol/signal. Les modèles basés sur la persistence topologique ont été testés sur une collection hétérogène de bases de données. Ces deux approches sont finalement combinées pour donner un troisième point de vue, qui a donné deux applications. Premièrement, on utilise l'alignement multiple des sequences, pour comparer plusieurs structures harmoniques et sémantiques déduites du signal audio, afin de visualiser et quantifier la propagation d’idée musicales entre artistes, genres et différentes époques. Ensuite on développe la théorie nécessaire pour comparer deux systèmes qui varient dans le temps, en représentant leurs caractéristiques géométriques comme des séries temporelles de diagrammes de persistence. / In this work, we suggest a collection of novel models for the representation of music. These models are endowed with two main features. First, they originate from a topological and geometrical inspiration; second, their low dimensionality allows to build simple and informative visualisations. We tackle the problem of music representation following three non-orthogonal directions. First, we propose an interpretation of counterpoint as a multivariate time series of partial permutation matrices, whose observations are characterised by a degree of complexity. After providing both a static and a dynamic representation of counterpoint, voice leadings are reinterpreted as a special class of partial singular braids, and their main features are visualised. Thereafter, we give a topological interpretation of the Tonnetz (a graph commonly used in computational musicology), whose vertices are deformed by both a harmonic and a consonance-oriented function. The shapes derived from these deformations are classified using the formalism of persistent homology. Thus, this novel representation of music is evaluated on a collection of heterogenous musical datasets. Finally, a combination of the two approaches is proposed. A model at the crossroad between the signal and symbolic analysis of music uses multiple sequences alignment to provide an encompassing, novel viewpoint on the musical inspiration transfer among compositions belonging to different artists, genres and time. Then, music is represented as a time series of topological fingerprints, allowing the comparison of pairs of time-varying shapes in both topological and musical terms.

Nous nous intéressons au développement de méthodes qui répondent aux difficultés posées par l’analyse des séries temporelles. Nos contributions se focalisent sur deux tâches : la prédiction de séries temporelles et la classification de séries temporelles. Notre première contribution présente une méthode de prédiction et de complétion de séries temporelles multivariées et relationnelles. Le but est d’être capable de prédire simultanément l’évolution d’un ensemble de séries temporelles reliées entre elles selon un graphe, ainsi que de compléter les valeurs manquantes dans ces séries (pouvant correspondre par exemple à une panne d’un capteur pendant un intervalle de temps donné). On se propose d’utiliser des techniques d’apprentissage de représentation pour prédire l’évolution des séries considérées tout en complétant les valeurs manquantes et prenant en compte les relations qu’il peut exister entre elles. Des extensions de ce modèle sont proposées et décrites : d’abord dans le cadre de la prédiction de séries temporelles hétérogènes puis dans le cas de la prédiction de séries temporelles avec une incertitude exprimée. Un modèle de prédiction de séries spatio-temporelles est ensuiteproposé, avec lequel les relations entre les différentes séries peuvent être exprimées de manière plus générale, et où ces dernières peuvent être apprises.Enfin, nous nous intéressons à la classification de séries temporelles. Un modèle d’apprentissage joint de métrique et de classification de séries est proposé et une comparaison expérimentale est menée. / This thesis deals with the development of time series analysis methods. Our contributions focus on two tasks: time series forecasting and classification. Our first contribution presents a method of prediction and completion of multivariate and relational time series. The aim is to be able to simultaneously predict the evolution of a group of time series connected to each other according to a graph, as well as to complete the missing values ​​in these series (which may correspond for example to a failure of a sensor during a given time interval). We propose to use representation learning techniques to forecast the evolution of the series while completing the missing values ​​and taking into account the relationships that may exist between them. Extensions of this model are proposed and described: first in the context of the prediction of heterogeneous time series and then in the case of the prediction of time series with an expressed uncertainty. A prediction model of spatio-temporal series is then proposed, in which the relations between the different series can be expressed more generally, and where these can be learned.Finally, we are interested in the classification of time series. A joint model of metric learning and time-series classification is proposed and an experimental comparison is conducted.

La convergence d 'une série de Fourier est toujours un thème d'actualité. Notre sujet d 'étude porte sur la convergence presque partout de la série de Fourier d 'une fonction· de carré sommable, démontrée par L. Carleson en 1966 en utilisant une décomposition de cette fonction. C. Feffermann, en 1973, utilise la décomposition de l'opérateur de Carleson. Puis, dans la deuxième moitié des années 1990, C. Thiele et M. T . Lacey reprennent cette décomposit ion en s 'appuyant sur la st ructure du plan t emps-fréquence. C'est dans cette optique qu'on va discuter de ce sujet , en tirant profit de la géométrie du plan temps-fréquence et de l'analyse des paquets d 'ondes.

Ce mémoire se consacre à l’analyse de la série télévisuelle Les Invincibles, diffusée sur les ondes de Radio-Canada entre 2005 et 2009. Il questionne la relation entre les trois récits mis en cause au sein de la télésérie, soit le récit principal qui est une fiction vraisemblable, le premier récit secondaire qui évoque la téléréalité et le deuxième récit secondaire aux allures de bande dessinée. Par le recours à l’intermédialité et l’intergénéricité, ce mémoire s’intéresse principalement aux emprunts de formes filmiques, télévisuelles et bédéiques et aux dimensions génériques dont la fiction, le documentaire et le fantastique qui sont associées à ces formes. En participant à la construction de l’histoire, cette tresse de récits témoigne des transformations que subissent les fictions télévisuelles québécoises actuelles et contribue ainsi à l’évolution du genre.