Cette thèse présente un couplage thermodynamique pour une modélisation approfondie du processus de soudure par ultrasons, surtout la soudure des thermoplastiques. Il s'agit de bien connaître le mode de fonctionnement des ensembles acoustiques réalisant la soudure ainsi que le comportement des matériaux à souder. En conséquence, les propriétés de ces matériaux, surtout celles nécessaires pour la modélisation par éléments finis, sont identifiées. Les paramètres mécaniques et thermiques sont mesurés en statique et ils sont comparés avec les résultats calcules par homogénéisation. Certains sont déterminés en dynamique, selon la fréquence et aussi en fonction de la température. Ceci est très utile parce que les matériaux travaillent à haute fréquence ultrasonique et que la température lors de la soudure change fortement. La machine de soudage par ultrasons (l'ensemble acoustique) a déjà été conçue et fabriquée. La modélisation EF en 3D avec Abaqus nous montre bien ses comportements modaux et vibratoires. Ensuite, les matériaux à souder sont introduits en modélisants le contact dynamique entre l'ensemble acoustique et la bande à souder. Ce modèle nous permet de déterminer le temps de contact et la compression dans les matériaux en fonction de la force de maintien. La dissipation d'énergie qui est engendrée par viscosité des matériaux à souder est calculée et introduite dans le modèle couplé thermomécanique. Le transfert de chaleur dans tout l'ensemble lors de la soudure est modélisé et il montre le champ de température, surtout la température à l'interface entre deux couches de matériaux. Cette modélisation est complétée en utilisant la méthode ALE (Arbitrary Lagrangian Eulerian) afin de tenir compte du mouvement d'avancement des bandes à souder

La réalisation d'une mémoire quantique pour la lumière met en jeu les aspects les plus fondamentaux de l'interaction matière-rayonnement. Pour capturer l'information quantique portée par la lumière, le matériau doit être capable de se maintenir dans un état de superposition quantique. Le temps de stockage est limité par la durée de vie de cet état, caractérisée par le temps de cohérence. Les premières expériences ont été réalisées dans des vapeurs atomiques froides, bien connues. Plus récemment, les ions de terres rares en matrice cristalline (REIC) ont attiré l'attention par leurs long temps de cohérence, associés à de larges bandes passantes d'interaction. Pour exploiter ces bonnes propriétés, des protocoles spécifiques ont été proposés. Nous nous sommes tournés vers un dérivé prometteur de l'écho de photon, le Peigne Atomique de Fréquences (AFC, proposé en 2008), fondé sur la transmission du champ incident à travers un profil d'absorption spectralement périodique. Les premiers chapitres de ce manuscrit présentent ce protocole et les travaux effectués durant cette thèse pour en améliorer l'efficacité (i.e. la probabilité de capter et de restituer l'information incidente), en augmenter la bande passante et la capacité de multiplexage et en mesurer le bruit. Les chapitres suivants présentent un nouveau protocole, proposé dans notre groupe durant cette thèse, et baptisé ROSE (Revival Of Silenced Echo). Ce protocole, très proche de l'écho de photon, a été démontré et caractérisé expérimentalement. Il semble très prometteur en termes d'efficacité, de bande passante et de bruit.

In this dissertation, the problem of learning from highly imbalanced data is studied. Imbalance data learning is of great importance and challenge in many real applications. Dealing with a minority class normally needs new concepts, observations and solutions in order to fully understand the underlying complicated models. We try to systematically review and solve this special learning task in this dissertation.We propose a new ensemble learning framework—Diversified Ensemble Classifiers for Imbal-anced Data Learning (DECIDL), based on the advantages of existing ensemble imbalanced learning strategies. Our framework combines three learning techniques: a) ensemble learning, b) artificial example generation, and c) diversity construction by reversely data re-labeling. As a meta-learner, DECIDL utilizes general supervised learning algorithms as base learners to build an ensemble committee. We create a standard benchmark data pool, which contains 30 highly skewed sets with diverse characteristics from different domains, in order to facilitate future research on imbalance data learning. We use this benchmark pool to evaluate and compare our DECIDL framework with several ensemble learning methods, namely under-bagging, over-bagging, SMOTE-bagging, and AdaBoost. Extensive experiments suggest that our DECIDL framework is comparable with other methods. The data sets, experiments and results provide a valuable knowledge base for future research on imbalance learning. We develop a simple but effective artificial example generation method for data balancing. Two new methods DBEG-ensemble and DECIDL-DBEG are then designed to improve the power of imbalance learning. Experiments show that these two methods are comparable to the state-of-the-art methods, e.g., GSVM-RU and SMOTE-bagging. Furthermore, we investigate learning on imbalanced data from a new angle—active learning. By combining active learning with the DECIDL framework, we show that the newly designed Active-DECIDL method is very effective for imbalance learning, suggesting the DECIDL framework is very robust and flexible.Lastly, we apply the proposed learning methods to a real-world bioinformatics problem—protein methylation prediction. Extensive computational results show that the DECIDL method does perform very well for the imbalanced data mining task. Importantly, the experimental results have confirmed our new contributions on this particular data learning problem.

This thesis is focused on the analysis and development of Ensembles of Neural Networks. An ensemble is a system in which a set of heterogeneous Artificial Neural Networks are generated in order to outperform the Single network based classifiers. However, this proposed thesis differs from others related to ensembles of neural networks [1, 2, 3, 4, 5, 6, 7] since it is organized as follows. In this thesis, firstly, an ensemble methods comparison has been introduced in order to provide a rank-based list of the best ensemble methods existing in the bibliography. This comparison has been split into two researches which represents two chapters of the thesis. Moreover, there is another important step related to the ensembles of neural networks which is how to combine the information provided by the neural networks in the ensemble. In the bibliography, there are some alternatives to apply in order to get an accurate combination of the information provided by the heterogeneous set of networks. For this reason, a combiner comparison has also been introduced in this thesis. Furthermore, Ensembles of Neural Networks is only a kind of Multiple Classifier System based on neural networks. However, there are other alternatives to generate MCS based on neural networks which are quite different to Ensembles. The most important systems are Stacked Generalization and Mixture of Experts. These two systems will be also analysed in this thesis and new alternatives are proposed. One of the results of the comparative research developed is a deep understanding of the field of ensembles. So new ensemble methods and combiners can be designed after analyzing the results provided by the research performed. Concretely, two new ensemble methods, a new ensemble methodology called Cross-Validated Boosting and two reordering algorithms are proposed in this thesis. The best overall results are obtained by the ensemble methods proposed. Finally, all the experiments done have been carried out on a common experimental setup. The experiments have been repeated ten times on nineteen different datasets from the UCI repository in order to validate the results. Moreover, the procedure applied to set up specific parameters is quite similar in all the experiments performed. It is important to conclude by remarking that the main contributions are: 1) An experimental setup to prepare the experiments which can be applied for further comparisons. 2) A guide to select the most appropriate methods to build and combine ensembles and multiple classifiers systems. 3) New methods proposed to build ensembles and other multiple classifier systems.

Depuis le début des années 1990, le milieu local et communautaire est de plus en plus invité (voire même forcé) à jouer un rôle accru en matière de gestion municipale. Notre étude cherche à déterminer si l'évolution des pratiques de participation publique du projet de redéveloppement Benny Farm reflète de nouvelles formes de gestion urbaine en aménagement, orientées vers des processus de concertation et des mécanismes administratifs qui s'apparentent à un modèle de gouvernance locale participative. Cette étude cherche à comprendre la dynamique d'ensemble plutôt qu'à mesurer les motivations des différents intervenants impliqués dans le processus de consultation publique de Benny Farm. Elle cherche à déterminer en quoi une approche comme la gouvernance locale participative offre des avantages face aux formes de participation plus traditionnelles comme la consultation publique en ce qui concerne l'élaboration de projets urbains à caractère sensible comme Benny Farm. Benny Farm constitue à cet égard un projet d'aménagement urbain particulier, où différentes formules participatives se sont succédées, dans le cadre d'un même projet, sous l'autorité de trois administrations municipales différentes: le RCM, Vision Montréal et l'UCIM. L'évolution de ces différentes formules sur une période de 15 ans, soit de 1991 à 2005, en fait un cas unique qui permet une réflexion sur les rapports qu'entretiennent gouvernance locale et participation. La méthodologie choisie est de type qualitatif. Deux principaux outils de collecte ont été utilisés afin d'obtenir des éléments de réponse à nos interrogations: l'entrevue semi-dirigée et l'analyse documentaire (mémoires, rapports, revue de presse). Finalement, notre analyse laisse supposée que la gouvernance peu se révéler comme un nouvel outil de planification local, conjuguant résolution de conflit et concertation.

Non-linear Bayesian estimation, or estimation of the state of a non-linear stochastic system from a set of indirect noisy measurements is a problem encountered in several fields of science. The particle filter and the ensemble Kalman filter are both used to get sub-optimal solutions of Bayesian inference problems, particularly for high-dimensional non-Gaussian and non-linear models. Both are essentially Monte Carlo techniques that compute their results using a set of estimated trajectories of the variable to be monitored. It has been shown that in a linear and Gaussian environment, solutions obtained from both these filters converge to the optimal solution obtained by the Kalman Filter. However, it is of interest to explore how the two filters compare to each other in basic methodology and construction, especially due to the similarity between them. In this work, we take up a specific problem of Bayesian inference in a restricted framework and compare analytically the results obtained from the particle filter and the ensemble Kalman filter. We show that for the chosen model, under certain assumptions, the two filters become methodologically analogous as the sample size goes to infinity.

Non-linear Bayesian estimation, or estimation of the state of a non-linear stochastic system from a set of indirect noisy measurements is a problem encountered in several fields of science. The particle filter and the ensemble Kalman filter are both used to get sub-optimal solutions of Bayesian inference problems, particularly for high-dimensional non-Gaussian and non-linear models. Both are essentially Monte Carlo techniques that compute their results using a set of estimated trajectories of the variable to be monitored. It has been shown that in a linear and Gaussian environment, solutions obtained from both these filters converge to the optimal solution obtained by the Kalman Filter. However, it is of interest to explore how the two filters compare to each other in basic methodology and construction, especially due to the similarity between them. In this work, we take up a specific problem of Bayesian inference in a restricted framework and compare analytically the results obtained from the particle filter and the ensemble Kalman filter. We show that for the chosen model, under certain assumptions, the two filters become methodologically analogous as the sample size goes to infinity.

To increase the trend of placing more importance on arts and humanity in the current wave of educational reform, Ministry of Education endeavors to promote the "one school, one art group" project to realize the education of the arts and humanity and to urge the necessity of interdisciplinary teaching in the art category, in combination with the "One person, one musical instrument" project to pursue a wider range of learning. In an environment that art subject and after class activities used to be neglected but now attract more attention, string orchestra is gradually being developed. The members of junior high school and elementary school string orchestras are mostly the basic members of senior high school orchestras and even the music majors in universities. The promotion of basic string education will be beneficial to a higher level of the string music. The first chapter mainly researches into the motivation, purpose, objects, scope and method of this study. The second chapter primarily focuses on the understanding of the development of string orchestras in other countries, and the approach of development, function and meaning of string orchestras in Taiwan. The third chapter introduces string orchestras in junior high school and elementary school, including the preparation of the string orchestra, the recruitment of members, the training and the teaching materials applied in an orchestra, and so on. The fourth chapter explores the current situation of the junior high school and elementary school string orchestras in Kaohsiung city, including current teaching conditions, practicing repertoires, teaching materials, the problems encountered, and so on. Conclusion and suggestions are proposed in the fifth chapter .The development of string orchestras in Taiwan have gradually matured with an effort from many forerunners and string teachers in recent years. As a music educator, I hope that more relevant information about the teaching situations in junior high school and elementary school string orchestras in Kaohsiung City can be acquired through this study and that I can make a contribution to string music.

There are about 1500 wind bands in Taiwan, which are separated in primary, junior high, senior high school, universities and communities etc. In Taiwan, the population who learn the wind instrument are more than one hundred thousand population, however, there are only five professional orchestras, including National Symphony Orchestra, Taipei Symphony Orchestra, National Taiwan Symphony Orchestra, Kaohsiung City Symphony Orchestra and Evergreen Symphony Orchestra, and no a professional wind band. Therefore, the purpose of this study is to understanding the availability and possibility to fund a Professional Wind Band in Taiwan via evaluating the market of musical performing and the status of audience involvement. The study methods contain conducting interviewing the key person of the wind band and designing a survey with the factors of Human Resource, Financial Plan, and Operation Status.

Determination of bio-material by using enzyme, fluorophore or metal-nanoparticles as markers is very important. Generally, gold-nanoparticles have been used frequently as marker for increasing the sensitivity in bio-chemical assay. In this research, gold-nanoparticles were used as marker for immunoassay, DNA sequence assay, and protein analysis. However, the size of gold-nanoparticles affects directly the results of electrochemical detection. For improving the sensitivity of electrochemical method, enlargement of gold-nanoparticles was used in this study. By electroless deposition, Au will be deposited on the surface of gold-nanoparticles. The electrochemical response will thus be increased substantially. In immunoassay and DNA sequence assay, traditional 96-wells microtiter plate was used for immobilizing antibody or oligonucleotide, and the gold-nanoparticles were marked subsequently base on the immunoreaction or protein reaction of streptavidin and biotin. After gold-nanoparticles were enlarged, they were dissolved and transferred to an electrochemical cell for square wave stripping voltammetry¡]SWSV¡^analysis. Under optimal experimental condition, dynamic range of 1 ~ 500 pg/mL and 0.52 ~ 1300 aM were found respectively for RIgG and Target DNA analysis, and a good linear relationship¡]R2 = 0.9975 and 0.9982¡^. The relative standard deviation¡]R.S.D.¡^ of blank were 2.8 % and 2.4 %¡]n = 11¡^for immunoassay and DNA assay, respectively. And the variance was 2.4 %¡]n = 9¡^and 2.4 %¡]n = 12¡^for immunoassay and DNA assay, respectively. The detection limit¡]based on S/N = 3¡^of RIgG and DNA were 0.25 pg/mL and 0.52 aM, respectively. They are very competitive compared with similar results reported in the literature. Additional, a gold nanoelectrode ensemble¡]GNEE¡^coupled microchip system was developed for bio-electrochemical analysis. Due to the difference in mobility of urea and urease were mixed and allowed the enzymatic reaction to proceed in microchannel. The enzymatic product NH4+ was determined by the coupled GNEE at the outlet of the channel. Another experiment of streptavidin conjugated gold-nanoparticles¡]streptavidin-Au¡^, reductant and gold-ion¡]Au3+¡^solution was be applied here, too. The product, NH4+ or Au3+ was passed through downstream of microchannel and detected by GNEE of electrochemical system. Satisfactory linear relationship¡]R2 = 0.9778 and 0.9657¡^were found from 0.1 mM to 50 mM for NH4+ and urea in the range of 0.02 mM to 5.0 mM, respectively. The other satisfactory linear relationship¡]R2 = 0.9842 and 0.9507¡^ were found between 3.75 mg/mL and 3.75 g/mL for Au3+ and streptavidin-Au in the range of 0.2 ng/mL to 100 ng/mL, respectively. Variances of 2.5 %¡]n = 6¡^was found for analysis of with the microchip system.