The Eastman Wind Ensemble is known as the pioneer ensemble of modern wind music in North America and abroad. Its founder and conductor, Frederick Fennell, was instrumental in facilitating the creation and performance of a large number of new works written for the specific instrumentation of the wind ensemble. Created in 1952, the EWE developed a new one-to-a-part instrumentation that could be varied based on the wishes of the composer. This change in instrumentation allowed for many more compositional choices when composing. The instrumentation was a dramatic shift from the densely populated ensembles that were standard in North America by 1952. The information on the EWE and Fennell is available at the Eastman School of Music’s Ruth Watanabe Archive. By comparing the repertory and instrumentation of the Eastman ensembles with other contemporary ensembles, Fennell’s revolutionary ideas are shown to be unique in the wind music community.

Syftet med undersökningen är att med avseende på skolans genusuppdrag studera musiklärares förhållningsätt i ensembleundervisning på gymnasieskolan. För att uppnå detta har kvalitativa, semistrukturerade intervjuer genomförts med lärare i gymnasiekursen ensemble, följt av diskursanalys på transkriberingar av dessa intervjuer. De teoretiska utgångspunkterna är genusteorier och diskursteorier. I resultatet framkommer lärarnas syn på genusbegreppet och genusuppdraget. Bland de genusdiskurser som presenteras finns synen på genus som rutin, som sidoeffekt på annat arbete och som medvetenhet. Gällande elevsyn visar sig diskurser om eleven som platstagare, som normbrytare och vikten av att eleven känner sig som en synbar individ. Slutledningarna, liksom diskussionskapitlet, rör sig kring genusarbetets framgångar, vilka faktorer som kan ha orsakat dessa, och hur arbetet förändrats på skolorna utifrån dessa framgångar. Genomgående har ett individfokus framkommit, där eleverna inte ses som en grupp, utan som enskilda individer, vilka alla måste mötas personligt. / The purpose of the following study is to examine the approach of music teachers to the charge of countering traditional gender roles in the subject ensemble in secondary high school. To achieve this qualitative, semi-structured interviews have been implemented with teachers in the secondary high school subject of ensemble, followed by discourse analyses of the transcriptions of said interviews. The theoretical basis for this study is made up of gender theories and discourse theory. The result presents a number of discourses regarding gender and pupils. Amongst these discourses can be found thoughts on gender work as routine, as a side effect of other things and as a constant awareness. Regarding pupils, there are discourses on pupils as space takers, as norm breakers and the importance of every pupil feeling like an observed individual. The conclusions, as well as the discussion chapter, brings up the success of gender work, what may have caused it, and how the daily work has come to change due to these successes. Throughout the study, a focus on the individual has presented itself, where pupils are not seen as a group, but as separate individuals, who must be met on a personal plain.

Ce travail porte sur l'estimation des incertitudes et la prévision de risques en qualité de l'air. Il consiste dans un premier temps à construire un ensemble de simulations de la qualité de l'air qui prend en compte toutes les incertitudes liées à la modélisation de la qualité de l'air. Des ensembles de simulations photochimiques à l'échelle continentale ou régionale sont générés automatiquement. Ensuite, les ensembles générés sont calibrés par une méthode d'optimisation combinatoire qui sélectionne un sous-ensemble représentatif de l'incertitude ou performant (fiabilité et résolution) pour des prévisions probabilistes. Ainsi, il est possible d'estimer et de prévoir des champs d'incertitude sur les concentrations d'ozone ou de dioxyde d'azote, ou encore d'améliorer la fiabilité des prévisions de dépassement de seuil. Cette approche est ensuite comparée avec la calibration d'un ensemble Monte Carlo. Ce dernier, moins dispersé, est moins représentatif de l'incertitude. Enfin, on a pu estimer la part des erreurs de mesure, de représentativité et de modélisation de la qualité de l'air / This work is about uncertainty estimation and risk prediction in air quality. Firstly, we need to build an ensemble of air quality simulations which can take into account all uncertainty sources related to air quality modeling. Ensembles of photochemical simulations at continental and regional scales are automatically built. Then, these generated ensemble are calibrated with a combinatorial optimization method. It selects a sub-ensemble which is representative of uncertainty or has good resolution and reliability of probabilistic forecasts. Thus, this work show that it is possible to estimate and forecast uncertainty fields related to ozone and nitrogen dioxide concentrations or to improve reliability related to the threshold exceedance prediction. This approach is compared with Monte Carlo ensemble calibration. This ensemble is less representative of uncertainty. Finally, we can estimate the part of the measure error, representativity error and modeling error in air quality

As características irrelevantes, presentes em bases de dados de diversos domínios, deterioram a acurácia de predição de classificadores induzidos por algoritmos de aprendizado de máquina. As bases de dados geradas por uma língua eletrônica são exemplos típicos onde a demasiada quantidade de características irrelevantes e redundantes prejudicam a acurácia dos classificadores induzidos. Para lidar com este problema, duas abordagens podem ser utilizadas. A primeira é a utilização de métodos para seleção de subconjuntos de características. A segunda abordagem é por meio de ensemble de classificadores. Um ensemble deve ser constituído por classificadores diversos e acurados. Uma forma efetiva para construção de ensembles de classificadores é por meio de seleção de características. A seleção de características para ensemble tem o objetivo adicional de encontrar subconjuntos de características que promovam acurácia e diversidade de predição nos classificadores do ensemble. Algoritmos genéticos são técnicas promissoras para seleção de características para ensemble. No entanto, a busca genética, assim como outras estratégias de busca, geralmente visam somente a construção do ensemble, permitindo que todas as características (relevantes, irrelevantes e redundantes) sejam utilizadas. Este trabalho apresenta uma abordagem baseada em algoritmos genéticos para construção de ensembles de redes neurais artificiais com um conjunto reduzido das características totais. Para melhorar a acurácia dos ensembles, duas abordagens diferenciadas para treinamento de redes neurais foram utilizadas. A primeira baseada na interrupção precoce do treinamento com o algoritmo back-propagation e a segunda baseada em otimização multi-objetivo. Os resultados obtidos comprovam a eficácia do algoritmo proposto para construção de ensembles de redes neurais acurados. Também foi constatada sua eficiência na redução das características totais, comprovando que o algoritmo proposto é capaz de construir um ensemble utilizando um conjunto reduzido de características. / The irrelevant features in databases of some domains spoil the accuracy of the classifiers induced by machine learning algorithms. Databases generated by an electronic tongue are examples where the huge quantity of irrelevant and redundant features spoils the accuracy of classifiers. There are basically two approaches to deal with this problem: feature subset selection and ensemble of classifiers. A good ensemble is composed by accurate and diverse classifiers. An effective way to construct ensembles of classifiers is to make it through feature selection. The ensemble feature selection has an additional objective: to find feature subsets to promote accuracy and diversity in the ensemble of classifiers. Genetic algorithms are promising techniques for ensemble feature selection. However, genetic search, as well as other search strategies, only aims the ensemble construction, allowing the selection of all features (relevant, irrelevant and redundant). This work proposes an approach based on genetic algorithm to construct ensembles of neural networks using a reduced feature subset of totality. Two approaches were used to train neural networks to improve the ensembles accuracy. The first is based on early stopping with back-propagation algorithm and the second is based on multi-objective optimization. The results show the effectiveness and accuracy of the proposed algorithm to construct ensembles of neural networks, and also, its efficiency in the reduction of total features was evidenced, proving its capacity for constructing an ensemble using a reduced feature subset.

La stéganographie est l'art de la communication secrète. L'objectif est de dissimuler un message secret dans un médium anodin de sorte qu'il soit indétectable. De nos jours, avec la généralisation d'Internet et l'apparition des supports numériques (fichiers audio, vidéos ou images), plusieurs philosophies de conception de schéma stéganographique ont été proposées. Parmi les méthodes actuelles appliquées aux images numériques naturelles, nous trouvons les méthodes d'insertion adaptative, dont le principe repose sur la modification du médium de couverture avec une garantie d'avoir un certain niveau de sécurité. Ces méthodes représentent une véritable avancée en stéganographie.Dans ce manuscrit, après avoir rappelé les concepts récents de stéganographie adaptative, nous présentons une procédure automatique et complète pour l'insertion adaptative de données secrètes dans des images numériques naturelles. L'approche proposée est une « méta-méthode » basée « oracle », appelée ASO (Adaptive Steganography by Oracle), qui permet de préserver à la fois la distribution de l'image de couverture et la distribution de la base d'images utilisée par l'émetteur. Notre approche permet d'obtenir des résultats nettement supérieurs aux méthodes actuelles de l'état de l'art, et est donc l'une, si ce n'est la meilleure approche du moment. Par ailleurs, nous définissons également un nouveau paradigme en stéganographie qui est la stéganographie par base, ainsi qu'une nouvelle mesure de sélection pour les images stéganographiées, permettant d'améliorer encore plus les performances de notre schéma d'insertion. Les différentes expérimentations, que nous avons effectuées sur des images réelles, ont confirmé la pertinence de cette nouvelle approche. / Steganography is the art of secret communication. The goal is to hide a secret message in an unsuspicious object in such a way that no one can detect it. Nowadays, with the Internet spread and the emergence of digital supports (audio files, videos, or images), several philosophies of designing steganographic methods were proposed. One of the most usual embedding methods used with real digital images is the adaptive embedding algorithms, which is based on the modification of the cover image with a guarantee of a certain security level. These methods represent an important progress in steganography.In this Ph.D. Thesis, we present a fully automated procedure for the adaptive embedding of secret data in digital images. For this, after recalling the recent concepts of adaptive steganography, we first introduce a clear formalism to define a new "meta-method" steganographic approach based on "oracle", whichwe called ASO (Adaptive Steganography by Oracle). Then, we define a new steganographic paradigm called "the steganography by database paradigm", and propose a new selection criterion to further enhance the security of the transmission phase of ASO. Experimental results show that our embedding approach ASO provides the highest level of steganographic security. It is then currently the best or one of the best approaches of the state of the art.

Mitt projekt handlar om att studera hur de olika saxofonstorlekarna samspelar i enensemble. Syftet med studien är att undersöka hur jag förändrar mitt saxofonspel närjag byter från barytonsaxofon till sopransaxofon i en saxofonkvartett. Minaforskningsfrågor är: Hur ändrar jag mitt spel och min roll när jag går från att spelabarytonsaxofon till att spela sopransaxofon? Vad beror dessa skillnader på? Ibakgrunden tas begrepp som är viktiga för ämnet upp såsom rytm, intonation,embouchyr, stöd, formella roller och informella roller. Studiens teoretiska utgångspunktutgörs av ett hermeneutiskt perspektiv. I metodkapitlet presenteras de metoder jaganvänt, nämligen videoobservation och loggbok. Slutligen presenteras resultatet av deanalyser som har gjorts av bytet från barytonsaxofon till sopransaxofon. Studien gårigenom alla de förändringar jag har gjort gällande intonation, tempo, rytm, frasering;embouchyr, luft, artikulation; hållning, rörelse, motorik samt ledarskap och roller. / My project is about studying how the different saxophone sizes interact in an ensemble.The purpose of the study is to investigate how I change the way I play the saxophonewhen switching from baritone saxophone to soprano saxophone in a saxophone quartet.My research questions are: How do I change the way I play the saxophone and my role inthe ensemble when I switch from playing baritone saxophone to playing sopranosaxophone? What are these differences due to? In the background chapter, concepts thatare important to the topic are presented such as rhythm, intonation, embouchure,support, formal roles and informal roles. The theoretical foundation of the study is ahermeneutic perspective. The methodology chapter presents the methods I used,namely video observation and logbook. Finally, the results of the analyzes made of theswitch from baritone saxophone to soprano saxophone are presented. The studypresents all the changes I have made regarding intonation, pace, rhythm, phrasing;embouchure, air, articulation; posture, movement, motor skills, leadership and roles.

Previsões probabilísticas de precipitação foram obtidas a partir de um conjunto de simulações pelo modelo WRF e utilizadas como condição de contorno no modelo hidrológico TopModel para previsão hidrometeorológica na bacia do Rio Iguaçu, no estado do Paraná. Nas simulações de cheias, durante o período de elevação do volume de precipitação, o erro médio aritmético do conjunto de previsões foi menor que cada um dos membros utilizados nesse conjunto, indicando melhor destreza do conjunto médio em relação a qualquer previsão determinística. Na dissipação dos sistemas precipitantes, alguns membros obtiveram resultados melhores que o conjunto médio e, em geral, as previsões são confluentes. As melhores previsões de precipitação com o WRF foram obtidas com as combinações de microfísica Lin e convecção de Kain Fritsch, microfísica WSM 5 e convecção de Kain Fritsch e simulações defasadas em 6 horas. As simulações inicializadas em horários mais próximos da ocorrência do fenômeno não garantiram uma melhoria na distribuição de precipitação na bacia. A avaliação do sistema de previsão por conjuntos pelo índice de Brier (IB) e seus termos demonstrou níveis suficientes de confiabilidade e destreza para ser utilizada na maioria dos eventos de precipitação sobre a bacia do rio Iguaçu. Os valores do IB estiveram entre 0,15 e 0,3 com picos isolados. Os valores obtidos para o termo de incerteza estiveram entre 0,1 e 0,25 indicando bons resultados visto que o desejável é o mais próximo de zero. Nos eventos de chuva, o termo de confiabilidade apresentou valores próximos a 0,2 no período da manhã e valores entre 0,3 e 0,4 no período da tarde, com um acréscimo no final da integração. O índice de acerto foi de 60 % a 90 % durante o período de integração (48 horas) para o conjunto médio de previsões e entre 50 a 80% para a previsão determinística. Em todos os horários de simulação o erro de fase foi maior que o erro de amplitude, possivelmente devido aos atrasos da propagação dos sistemas precipitantes e aos efeitos de ajuste das condições físicas iniciais da atmosfera. Os erros de fase e amplitude foram menores na previsão probabilística em todo o período de integração. Assim como na previsão de precipitação, nas simulações de vazão o erro de fase foi maior que o erro de amplitude, indicando que o atraso nas previsões de variação da vazão ainda é o um desafio na previsão hidrometeorológica. Observou-se que o modelo hidrológico é bastante sensível a previsão de precipitação e, portanto, a melhoria das previsões de vazão é diretamente proporcional a diminuição dos erros nas previsões de precipitação. / Probabilistic forecast of precipitation from WRF model simulations was used as input in hydrological TopModel for streamlines forecast in Iguaçu Basin, Parana, southern Brazil. The arithmetic error of precipitation ensemble forecast was smaller than each individual member forecast error in the streamflow increase stage. It means the use of ensemble forecast was better than any deterministic forecast. But when the streamflow decreases, the results are confluent and some individual member forecast was better than ensemble. Simulations using Lin microphysical parameterization and Kain Fritsch, WSM 5 and Kain Fritsch and 6h lagged obtained the better results of precipitation over the basin. The use of runs with initial conditions near the precipitation time did not guarantee better results in the distribution of precipitation on the basin. The Brier Score (BS) of the ensemble system demonstrated that the system is very skillful with values between 0.15 and 0.3. Both uncertainty and reliability terms of BS, 0.1 0.25 and 0.2- 0.4, respectively, were encouraging for use hourly ensemble forecast of precipitation on the watershed. Ensemble forecast provide high values of hit scores (0.6 to 0.9) than deterministic forecast (0.5 to 0.8) at all period of integration. Due the delay in the forecasts of the precipitation systems, the phase error is predominant over amplitude during all time. Both errors were reduced using the ensemble forecasts. The phase errors in hydrological were greater than amplitude such as precipitation forecasts. Thus, for increase streamflow forecast it should reduced the errors in QPF forecasts.

As características irrelevantes, presentes em bases de dados de diversos domínios, deterioram a acurácia de predição de classificadores induzidos por algoritmos de aprendizado de máquina. As bases de dados geradas por uma língua eletrônica são exemplos típicos onde a demasiada quantidade de características irrelevantes e redundantes prejudicam a acurácia dos classificadores induzidos. Para lidar com este problema, duas abordagens podem ser utilizadas. A primeira é a utilização de métodos para seleção de subconjuntos de características. A segunda abordagem é por meio de ensemble de classificadores. Um ensemble deve ser constituído por classificadores diversos e acurados. Uma forma efetiva para construção de ensembles de classificadores é por meio de seleção de características. A seleção de características para ensemble tem o objetivo adicional de encontrar subconjuntos de características que promovam acurácia e diversidade de predição nos classificadores do ensemble. Algoritmos genéticos são técnicas promissoras para seleção de características para ensemble. No entanto, a busca genética, assim como outras estratégias de busca, geralmente visam somente a construção do ensemble, permitindo que todas as características (relevantes, irrelevantes e redundantes) sejam utilizadas. Este trabalho apresenta uma abordagem baseada em algoritmos genéticos para construção de ensembles de redes neurais artificiais com um conjunto reduzido das características totais. Para melhorar a acurácia dos ensembles, duas abordagens diferenciadas para treinamento de redes neurais foram utilizadas. A primeira baseada na interrupção precoce do treinamento com o algoritmo back-propagation e a segunda baseada em otimização multi-objetivo. Os resultados obtidos comprovam a eficácia do algoritmo proposto para construção de ensembles de redes neurais acurados. Também foi constatada sua eficiência na redução das características totais, comprovando que o algoritmo proposto é capaz de construir um ensemble utilizando um conjunto reduzido de características. / The irrelevant features in databases of some domains spoil the accuracy of the classifiers induced by machine learning algorithms. Databases generated by an electronic tongue are examples where the huge quantity of irrelevant and redundant features spoils the accuracy of classifiers. There are basically two approaches to deal with this problem: feature subset selection and ensemble of classifiers. A good ensemble is composed by accurate and diverse classifiers. An effective way to construct ensembles of classifiers is to make it through feature selection. The ensemble feature selection has an additional objective: to find feature subsets to promote accuracy and diversity in the ensemble of classifiers. Genetic algorithms are promising techniques for ensemble feature selection. However, genetic search, as well as other search strategies, only aims the ensemble construction, allowing the selection of all features (relevant, irrelevant and redundant). This work proposes an approach based on genetic algorithm to construct ensembles of neural networks using a reduced feature subset of totality. Two approaches were used to train neural networks to improve the ensembles accuracy. The first is based on early stopping with back-propagation algorithm and the second is based on multi-objective optimization. The results show the effectiveness and accuracy of the proposed algorithm to construct ensembles of neural networks, and also, its efficiency in the reduction of total features was evidenced, proving its capacity for constructing an ensemble using a reduced feature subset.

Les méthodes ensemblistes constituent un sujet de recherche très populaire au cours de la dernière décennie. Leur succès découle en grande partie de leurs solutions attrayantes pour résoudre différents problèmes d'apprentissage intéressants parmi lesquels l'amélioration de l'exactitude d'une prédiction, la sélection de variables, l'apprentissage de métrique, le passage à l'échelle d'algorithmes inductifs, l'apprentissage de multiples jeux de données physiques distribués, l'apprentissage de flux de données soumis à une dérive conceptuelle, etc... Dans cette thèse nous allons dans un premier temps présenter une comparaison empirique approfondie de 19 algorithmes ensemblistes d'apprentissage supervisé proposé dans la littérature sur différents jeux de données de référence. Non seulement nous allons comparer leurs performances selon des métriques standards de performances (Exactitude, AUC, RMS) mais également nous analyserons leur diagrammes kappa-erreur, la calibration et les propriétés biais-variance. Nous allons aborder ensuite la problématique d'amélioration des ensembles de modèles par la sélection dynamique d'ensembles (dynamic ensemble selection, DES). La sélection dynamique est un sous-domaine de l'apprentissage ensembliste où pour une donnée d'entrée x, le meilleur sous-ensemble en terme de taux de réussite est sélectionné dynamiquement. L'idée derrière les approches DES est que différents modèles ont différentes zones de compétence dans l'espace des instances. La plupart des méthodes proposées estime l'importance individuelle de chaque classifieur faible au sein d'une zone de compétence habituellement déterminée par les plus proches voisins dans un espace euclidien. Nous proposons et étudions dans cette thèse deux nouvelles approches DES. La première nommée ST-DES est conçue pour les ensembles de modèles à base d'arbres de décision. Cette méthode sélectionne via une métrique supervisée interne à l'arbre, idée motivée par le problème de la malédiction de la dimensionnalité : pour les jeux de données avec un grand nombre de variables, les métriques usuelles telle la distance euclidienne sont moins pertinentes. La seconde approche, PCC-DES, formule la problématique DES en une tâche d'apprentissage multi-label avec une fonction coût spécifique. Ici chaque label correspond à un classifieur et une base multi-label d'entraînement est constituée sur l'habilité de chaque classifieur de classer chaque instance du jeu de données d'origine. Cela nous permet d'exploiter des récentes avancées dans le domaine de l'apprentissage multi-label. PCC-DES peut être utilisé pour les approches ensemblistes homogènes et également hétérogènes. Son avantage est de prendre en compte explicitement les corrélations entre les prédictions des classifieurs. Ces algorithmes sont testés sur un éventail de jeux de données de référence et les résultats démontrent leur efficacité faces aux dernières alternatives de l'état de l'art / Ensemble methods has been a very popular research topic during the last decade. Their success arises largely from the fact that they offer an appealing solution to several interesting learning problems, such as improving prediction accuracy, feature selection, metric learning, scaling inductive algorithms to large databases, learning from multiple physically distributed data sets, learning from concept-drifting data streams etc. In this thesis, we first present an extensive empirical comparison between nineteen prototypical supervised ensemble learning algorithms, that have been proposed in the literature, on various benchmark data sets. We not only compare their performance in terms of standard performance metrics (Accuracy, AUC, RMS) but we also analyze their kappa-error diagrams, calibration and bias-variance properties. We then address the problem of improving the performances of ensemble learning approaches with dynamic ensemble selection (DES). Dynamic pruning is the problem of finding given an input x, a subset of models among the ensemble that achieves the best possible prediction accuracy. The idea behind DES approaches is that different models have different areas of expertise in the instance space. Most methods proposed for this purpose estimate the individual relevance of the base classifiers within a local region of competence usually given by the nearest neighbours in the euclidean space. We propose and discuss two novel DES approaches. The first, called ST-DES, is designed for decision tree based ensemble models. This method prunes the trees using an internal supervised tree-based metric; it is motivated by the fact that in high dimensional data sets, usual metrics like euclidean distance suffer from the curse of dimensionality. The second approach, called PCC-DES, formulates the DES problem as a multi-label learning task with a specific loss function. Labels correspond to the base classifiers and multi-label training examples are formed based on the ability of each classifier to correctly classify each original training example. This allows us to take advantage of recent advances in the area of multi-label learning. PCC-DES works on homogeneous and heterogeneous ensembles as well. Its advantage is to explicitly capture the dependencies between the classifiers predictions. These algorithms are tested on a variety of benchmark data sets and the results demonstrate their effectiveness against competitive state-of-the-art alternatives

The Eastman Wind Ensemble is known as the pioneer ensemble of modern wind music in North America and abroad. Its founder and conductor, Frederick Fennell, was instrumental in facilitating the creation and performance of a large number of new works written for the specific instrumentation of the wind ensemble. Created in 1952, the EWE developed a new one-to-a-part instrumentation that could be varied based on the wishes of the composer. This change in instrumentation allowed for many more compositional choices when composing. The instrumentation was a dramatic shift from the densely populated ensembles that were standard in North America by 1952. The information on the EWE and Fennell is available at the Eastman School of Music’s Ruth Watanabe Archive. By comparing the repertory and instrumentation of the Eastman ensembles with other contemporary ensembles, Fennell’s revolutionary ideas are shown to be unique in the wind music community.