Scientific applications are large, complex, irregular, and computationally intensive and are characterized by data parallel loops. The prevalence of independent iterations in these loops, makes parallel computing as the natural choice for solving these applications. The computational requirements of these problems vary due to variations in problem, algorithmic and systemic characteristics during parallelization, leading to performance degradation. Considerable amount of research has been dedicated to the development of dynamic scheduling techniques based on probabilistic analysis to address these predictable and unpredictable factors that lead to severe load imbalance. The mathematical foundations of these scheduling algorithms have been previously developed and published in the literature. These techniques have been successfully integrated into scientific applications as well as into runtime systems. Recently, efforts have also been directed to integrate these techniques into dynamic load balancing libraries for scientific applications. The optimal scheduling algorithm to load balance a specific scientific application in a dynamic parallel computing environment is very difficult without the exhaustive testing of all the scheduling techniques. This is a time consuming process, and therefore, there is a need for developing an automatic mechanism for the selection of dynamic scheduling algorithms. In recent years, extensive work has been dedicated to the development of reinforcement learning and some of its techniques have addressed load-balancing problems. However, they do not cover a number of aspects regarding the performance of scientific applications. First, these previously developed techniques address the load balancing problem only at a coarse granularity level (for example, job scheduling), and the reinforcement learning techniques used for load balancing are based on learning from trained datasets which are obtained prior to the execution of the application. Moreover, scientific applications contain parameters whose variations are so irregular that the use of training sets would not be able to accurately capture the entire spectrum of possible characteristics. Finally, algorithm selection using reinforcement learning has only been used for simple sequential problems. This thesis addresses these limitations and provides a novel integrated approach for automating the selection of dynamic scheduling algorithms at a finer granularity level to improve the performance of scientific applications using reinforcement learning. This integrated approach will experimentally be tested on a scientific application that involves a large number of time steps: The Quantum Trajectory Method (QTM). A qualitative and quantitative analysis of the effectiveness of this novel approach will be presented to underscore the significance of its use in improving the performance of large-scale scientific applications.

During flights with manned or unmanned aircraft, continuous recording can result in avery high number of images to analyze and evaluate. To simplify image analysis and tominimize data link usage, appropriate images should be suggested for transfer and furtheranalysis. This thesis investigates features used for selection of images worthy of furtheranalysis using machine learning. The selection is done based on the criteria of havinggood quality, salient content and being unique compared to the other selected images.The investigation is approached by implementing two binary classifications, one regardingcontent and one regarding quality. The classifications are made using support vectormachines. For each of the classifications three feature extraction methods are performedand the results are compared against each other. The feature extraction methods used arehistograms of oriented gradients, features from the discrete cosine transform domain andfeatures extracted from a pre-trained convolutional neural network. The images classifiedas both good and salient are then clustered based on similarity measures retrieved usingcolor coherence vectors. One image from each cluster is retrieved and those are the resultingimages from the image selection. The performance of the selection is evaluated usingthe measures precision, recall and accuracy. The investigation showed that using featuresextracted from the discrete cosine transform provided the best results for the quality classification.For the content classification, features extracted from a convolutional neuralnetwork provided the best results. The similarity retrieval showed to be the weakest partand the entire system together provides an average accuracy of 83.99%.

O presente trabalho consiste na investigação das características da imagem mamográfica digitalizada para definir automaticamente parâmetros de processamento em um esquema de diagnóstico auxiliado por computador (CAD) para mamografia, com o objetivo de se obter o melhor desempenho possível. Além disso, com base na aplicação dos resultados dessa primeira investigação, propõe-se também uma técnica de redução dos índices de falso-positivo em esquemas CAD visando à redução do número de biópsias desnecessárias. Para a definição automática dos parâmetros de processamento nas técnicas de detecção de microcalcificações e nódulos, foram extraídas algumas características das imagens, como desvio padrão, terceiro momento e o limiar de binarização. Utilizando o método de automatização proposto, observou-se um aumento de 20% no desempenho do esquema CAD (Az da curva ROC) em relação ao método não automatizado com parâmetro fixo. Para que fosse possível o processamento da imagem mamográfica inteira pelo esquema CAD e as técnicas desenvolvidas, foi desenvolvida também uma técnica para seleção automática de regiões de interesses, que recorta partes relevantes da mama para a segmentação. O índice de falsos positivos foi tratado por técnica específica desenvolvida com base na comparação das duas incidências típicas do exame mamográfico que, juntamente com a avaliação automática da imagem no pré-processamento para detecção de microcalcificações produziu uma redução significativa de 86% daquela taxa em relação ao procedimento de parâmetro fixo. / This present work consists on the investigation of mammographic image characteristics for automatic determination of image processing parameters for a mammography computer aided diagnosis scheme (CAD) in order to get optimal performance. Additionally, using the results obtained on this first investigation, it was also developed a new technique for the reduction of false-positive rates on CAD projects, which can result on the reduction of the number of unnecessary biopsies. For the automatic definition of the image processing parameters for the techniques of detection of microcalcifications and nodules, some image characteristics had been extracted, as standard deviation, third momentum and the thresholding value. Using the proposed automatization method it was reported an increase of 20% in the CAD performance (evaluated determining the ROC curve) in comparison to the non-automatic method (fixed parameter). Besides, for CAD schemes it is necessary to process the entire mammographic image. Thus, it was also developed a technique for automatic selection of regions of interests in the mammogram, which extracts better regions from breast image for further segmentation. False-positives rates was treated by a specific technique based on the comparison of the two typical incidences of mammographic examination that together with the automatic parameter determination method for microcalcification detection produced a significant reduction of 86% of that rate in relation to the procedure that uses fixed parameter.

O presente trabalho consiste na investigação das características da imagem mamográfica digitalizada para definir automaticamente parâmetros de processamento em um esquema de diagnóstico auxiliado por computador (CAD) para mamografia, com o objetivo de se obter o melhor desempenho possível. Além disso, com base na aplicação dos resultados dessa primeira investigação, propõe-se também uma técnica de redução dos índices de falso-positivo em esquemas CAD visando à redução do número de biópsias desnecessárias. Para a definição automática dos parâmetros de processamento nas técnicas de detecção de microcalcificações e nódulos, foram extraídas algumas características das imagens, como desvio padrão, terceiro momento e o limiar de binarização. Utilizando o método de automatização proposto, observou-se um aumento de 20% no desempenho do esquema CAD (Az da curva ROC) em relação ao método não automatizado com parâmetro fixo. Para que fosse possível o processamento da imagem mamográfica inteira pelo esquema CAD e as técnicas desenvolvidas, foi desenvolvida também uma técnica para seleção automática de regiões de interesses, que recorta partes relevantes da mama para a segmentação. O índice de falsos positivos foi tratado por técnica específica desenvolvida com base na comparação das duas incidências típicas do exame mamográfico que, juntamente com a avaliação automática da imagem no pré-processamento para detecção de microcalcificações produziu uma redução significativa de 86% daquela taxa em relação ao procedimento de parâmetro fixo. / This present work consists on the investigation of mammographic image characteristics for automatic determination of image processing parameters for a mammography computer aided diagnosis scheme (CAD) in order to get optimal performance. Additionally, using the results obtained on this first investigation, it was also developed a new technique for the reduction of false-positive rates on CAD projects, which can result on the reduction of the number of unnecessary biopsies. For the automatic definition of the image processing parameters for the techniques of detection of microcalcifications and nodules, some image characteristics had been extracted, as standard deviation, third momentum and the thresholding value. Using the proposed automatization method it was reported an increase of 20% in the CAD performance (evaluated determining the ROC curve) in comparison to the non-automatic method (fixed parameter). Besides, for CAD schemes it is necessary to process the entire mammographic image. Thus, it was also developed a technique for automatic selection of regions of interests in the mammogram, which extracts better regions from breast image for further segmentation. False-positives rates was treated by a specific technique based on the comparison of the two typical incidences of mammographic examination that together with the automatic parameter determination method for microcalcification detection produced a significant reduction of 86% of that rate in relation to the procedure that uses fixed parameter.

Lors de la validation des modèles de séries chronologiques, une hypothèse qui peut s'avérer importante porte sur la loi des données. L'approche préconisée dans ce mémoire utilise les tests lisses d'ajustement. Ce mémoire apporte des développements théoriques et empiriques des tests lisses pour les modèles autorégressifs moyennes mobiles (ARMA) vectoriels. Dans des travaux précédents, Ducharme et Lafaye de Micheaux (2004) ont développé des tests lisses d'ajustement reposant sur les résidus des modèles ARMA univariés. Tagne Tatsinkou (2016) a généralisé les travaux dans le cadre des modèles ARMA vectoriels (VARMA), qui s'avèrent potentiellement utiles dans les applications avec données réelles. Des considérations particulières au cas multivarié, telles que les paramétrisations structurées dans les modèles VARMA sont abordées. Les travaux de Tagne Tatsinkou (2016) sont complétés selon les angles théoriques et des études de simulations additionnelles sont considérées. Les nouveaux tests lisses reposent sur des familles de polynômes orthogonaux. Dans cette étude, une attention particulière est accordée aux familles de Legendre et d'Hermite. La contribution théorique majeure est une preuve complète que la statistique de test est invariante aux transformations linéaires affines lorsque la famille d'Hermite est adoptée. Les résultats de Tagne Tatsinkou (2016) représentent une première étape importante, mais ils sont incomplets quant à l'utilisation des résidus du modèle. Les tests proposés reposent sur une famille de densités sous les hypothèses alternatives d'ordre k. La sélection automatique de l'ordre maximal, basée sur les résultats de Ledwina (1994), est discutée. La sélection automatique est également implantée dans nos études de simulations. Nos études de simulations incluent des modèles bivariés et un modèle trivarié. Dans une étude de niveaux, on constate la bonne performance des tests lisses. Dans une étude de puissance, plusieurs compétiteurs ont été considérés. Il est trouvé que les tests lisses affichent des propriétés intéressantes de puissance lorsque les données proviennent de modèles VARMA avec des innovations dans la classe de lois normales contaminées. / When validating time series models, the distribution of the observations represents a potentially important assumption. In this Master's Thesis, the advocated approach uses smooth goodness-of-fit test statistics. This research provides theoretical and empirical developments of the smooth goodness of fit tests for vector autoregressive moving average models (VARMA). In previous work, Ducharme and Lafaye de Micheaux (2004) developed smooth goodness-of-fit tests designed for the residuals of univariate ARMA models. Later, Tagne Tatsinkou (2016) generalized the work within the framework of vector ARMA (VARMA) models, which prove to be potentially useful in real applications. Structured parameterizations, which are considerations specific to the multivariate case, are discussed. The works of Tagne Tatsinkou (2016) are completed, according to theoretical angles, and additional simulation studies are also considered. The new smooth tests are based on families of orthogonal polynomials. In this study, special attention is given to Legendre's family and Hermite's family. The major theoretical contribution in this work is a complete proof that the test statistic is invariant to linear affine transformations when the Hermite family is adopted. The results of Tagne Tatsinkou (2016) represent an important first step, but they were incomplete with respect to the use of the model residuals. The proposed tests are based on a family of densities under alternative hypotheses of order k. A data driven method to choose the maximal order, based on the results of Ledwina (1994), is discussed. In our simulation studies, the automatic selection is also implemented. Our simulation studies include bivariate models and a trivariate model. In the level study, we can appreciate the good performance of the smooth tests. In the power study, several competitors were considered. We found that the smooth tests displayed interesting power properties when the data came from VARMA models with innovations in the class of contaminated normal distributions.