Cada imagem pode ser representada como uma combinação de diversas características, como por exemplo o histograma de intensidades de cor ou propriedades de textura da imagem. Essas características compõem um vetor multidimensional que representa a imagem. É comum esse vetor ser dado como entrada para um método de classificação de padrões que, após aprender por meio de diversos exemplos, pode gerar um modelo de decisão. Estudos sugerem evidências de que a preparação das imagens-- por meio da especificação cuidadosa da aquisição, pré-processamento e segmentação-- pode impactar significativamente a classificação. Além da falta de tratamento das imagens antes da extração de características, o desbalanceamento de classes também se apresenta como um obstáculo para que a classificação seja satisfatória. Imagens possuem características que podem ser exploradas para melhorar a descrição dos objetos de interesse e, portanto, sua classificação. Entre as possibilidades de melhorias estão: a redução do número de intensidades das imagens antes da extração de características ao invés de métodos de quantização no vetor já extraído; e a geração de imagens a partir das originais, de forma a promover o balanceamento de bases de dados cujo número de exemplos de cada classe é desbalanceado. Portanto, a proposta desta dissertação é melhorar a classificação de imagens utilizando métodos de processamento de imagens antes da extração de características. Especificamente, busca analisar a influência do balanceamento de bases de dados e da quantização na classificação. Este estudo analisa ainda a visualização do espaço de características após os métodos de geração artificial de imagens e de interpolação das características extraídas das imagens originais (SMOTE), comparando como espaço original. A ênfase dessa visualização se dá na observação da importância do rebalanceamento das classes. Os resultados obtidos indicam que a quantização simplifica as imagens antes da extração de características e posterior redução de dimensionalidade, produzindo vetores mais compactos; e que o rebalanceamento de classes de imagens através da geração de imagens artificiais pode melhorar a classificação da base de imagens, em relação à classificação original e ao uso de métodos no espaço de características já extraídas. / Each image can be represented by a combination of several features like color frequency and texture properties. Those features compose a multidimensional vector, which represents the original image. Commonly this vector is given as an input to a classification method that can learn from examplesand build a decision model. The literature suggests that image preparation steps like acute acquisition, preprocessing and segmentation can positively impact such classification. Besides that, class unbalancing is also a barrier to achieve good classification accuracy. Some features and methods can be explored to improveobjects\' description, thus their classification. Possible suggestions include: reducing colors number before feature extraction instead of applying quantization methods to raw vectors already extracted; and generating synthetic images from original ones, to balance the number of samples in an uneven data set. We propose to improve image classification using image processing methods before feature extraction. Specifically we want to analyze the influence of both balancing and quantization methods while applied to datasets in a classification routine. This research also analyses the visualization of feature space after the artificial image generation and feature interpolation (SMOTE), against to original space. Such visualization is used because it allows us to know how important is the rebalacing method. The results show that quantization simplifies imagesby producing compacted vectors before feature extraction and dimensionality reduction; and that using artificial generation to rebalance image datasets can improve classification, when compared to the original one and to applying methods on the already extracted feature vectors.

Em aplicações de sensoriamento remoto, há diversos problemas nos quais há conhecimento predominante sobre uma categoria ou classe alvo, e pouco conhecimento sobre as demais categorias. Nesses casos, o treinamento de um classificador é prejudicado pelo desbalanceamento de classes. Assim, o estudo de características visuais para se definir o melhor subespaço de características pode ser uma alternativa viável para melhorar o desempenho dos classificadores. O uso de abordagens baseadas em detecção de anomalias também pode auxiliar por meio da modelagem da classe normal (comumente majoritária) enquanto todas as outras classes são consideradas como anomalias. Este estudo apresentou uma base de imagens de sensoriamento remoto, cuja aplicação é identificar entre regiões de cobertura vegetal e regiões de não cobertura vegetal. Para solucionar o problema de desbalanceamento entre as classes, foram realizados estudos das características visuais a fim de definir qual o conjunto de atributos que melhor representa os dados. Também foi proposta a criação de um pipeline para se tratar bases desbalanceadas de cobertura vegetal. Este pipeline fez uso de técnicas de seleção de características e aprendizado ativo. A análise de características apresentou que o subespaço usando o extrator BIC com o índice de vegetação ExG foi o que melhor distinguiu os dados. Além disso, a técnica de ordenação proposta mostrou bom desempenho com poucas dimensões. O aprendizado ativo também ajudou na criação de um modelo melhor, com resultados comparáveis com as melhores características visuais. / In remote sensing applications, there are several problems in which there is predominant knowledge about a target category or class, and little knowledge of the other categories. In such cases, the training of a classifier is hampered by the class imbalance. Thus, the study of visual characteristics to determine the best subspace characteristics may be a feasible alternative to improve the performance of classifiers. The use of anomaly detection-based approaches can also help through the normal class modeling (usually the major class) while considering all other classes as anomalies. This study presents a remote sensing image dataset, whose application is to classify regions of the image into vegetation coverage (related to plantation) and non-vegetation coverage. To solve the class imbalance problem, studies were conducted using several visual characteristics in order to define the set of attributes that best represent the data. A pipeline that deals with the vegetation classification problem and its class imbalance issues is also proposed. This pipeline made use of feature selection techniques and active learning. The visual features analysis showed that a subspace using the BIC extractor with EXG vegetation index was the best to distinguished the data. Also, and the proposed sorting-based feature selection achieved good results with a low dimensional subspaces. Furthermore, the active learning helped creating a better model, with results comparable with the best visual features.

No description available.

Community detection in graphs has been an important research topic for many fields. The aim of community detection is to extract from graphs those groups of nodes that present more connections between them than with the rest of the network. Detecting such groups at different scales can help understanding the global behaviour of the system. However, recent studies have shown that realworld graphs follow power-law distributions for degree and community sizes. Specifically, these graphs present many small communities but just a few large ones. This unbalanced community size distribution poses a great challenge for community detection algorithms.Most of the existing methods are based on global approaches that require information about the network to be processed as a whole. Thus, those techniques can not be applied when the graph is too big to fit into one single machine, or in distributed setting when the graph is partitioned among multiple machines. To solve this limitations, a completely decentralized community detection algorithm is presented. It is based on diffusion, following a vertex-centric approach that allows each node to decide the diffusion rates based on local information. It adds as well a mechanism for controlling the diffusion speed through a customizable function.We evaluate the algorithm with a variety of graphs with different levels of imbalance and community structures. Our algorithm is able to detect (almost) perfectly the communities when the imbalance between community sizes is not extreme. We show as well how the sizes of the detected communities can be controlled by the diffusion strategy, allowing for better detection of finer or coarser resolutions in hierarchical graphs. The algorithm is also compared to other two well-known existing methods, achieving similar results in most of the cases though with a higher computation time. / Gemenskap detektering i grafer har varit ett viktigt forsknings ämne förmånga områden. Gemenskapsdetekterings syftet är att extrahera ur grafernade grupper av noder som har mer kopplingar mellan varandra än med restenav nätverket. Att upptäcka sådana grupper i olika skaler kan hjälpa till att förstå systemets globala beteende. Däremot har nyliga studier visat att verkliga grafers grad och gemenskap storlek följer en potenslagen fördelning. Specifikt,dessa grafer uppvisar många små gemenskaper men bara några stora. Denhär obalanserade gemenskaps storleks fördelningen utgör en stor utmaning för gemenskapsdetekterings algoritmer.De flesta av de befintliga metoderna är baserade på globala tillvägagångssätt som kräver att information om nätverket behandlas som helhet. Således kan dessa tekniker inte tillämpas när grafen är för stor för att passa in i en enda maskin, eller på distribuerat sätt när grafen är uppdelad bland flera maskiner. För att lösa dessa begränsningar, uppvisas en helt decentraliserad gemenskapsdetekterings algoritm.Denär baserad pådiffusion som följer en vertex-centrerad tillvägagångssätt.Varje node valder diffusionshastigheten baserad på lokal information. Deninnehåller även en mekanism som kontrollerar diffusionens hastighet genom en anpassningsbar funktion.Vi utvärderar algoritmen genom flera olika grafer med olika nivåer av obalans och gemenskaps strukurer. Vår algoritm kan (nästan) felfritt upptäcka gemenskaper där obalansen mellan dem inte är för stor. Vi visar även hur storlekenpå de hittade gemenskaperna kan kontrolleras genom diffusions strategin, somtillåter bättre uptäckt av finare eller grövre resolution av hierarkiska grafer. Algoritmen kan också jämföras med två befintliga, välkända metoder, vilka ger liknande resultat i de flesta fallen men tar längre tid att genomföra.

No description available.

Gaussian graphical model has been a popular tool to investigate conditional dependency between random variables by estimating sparse precision matrices. The estimated precision matrices could be mapped into networks for visualization. For related but different classes, jointly estimating networks by taking advantage of common structure across classes can help us better estimate conditional dependencies among variables. Furthermore, there may exist multilevel structure among variables; some variables are considered as higher level variables and others are nested in these higher level variables, which are called lower level variables. In this dissertation, we made several contributions to the area of joint estimation of Gaussian graphical models across heterogeneous classes: the first is to propose a joint estimation method for estimating Gaussian graphical models across unbalanced multi-classes, whereas the second considers multilevel variable information during the joint estimation procedure and simultaneously estimates higher level network and lower level network. For the first project, we consider the problem of jointly estimating Gaussian graphical models across unbalanced multi-class. Most existing methods require equal or similar sample size among classes. However, many real applications do not have similar sample sizes. Hence, in this dissertation, we propose the joint adaptive graphical lasso, a weighted L1 penalized approach, for unbalanced multi-class problems. Our joint adaptive graphical lasso approach combines information across classes so that their common characteristics can be shared during the estimation process. We also introduce regularization into the adaptive term so that the unbalancedness of data is taken into account. Simulation studies show that our approach performs better than existing methods in terms of false positive rate, accuracy, Mathews correlation coefficient, and false discovery rate. We demonstrate the advantage of our approach using liver cancer data set. For the second one, we propose a method to jointly estimate the multilevel Gaussian graphical models across multiple classes. Currently, methods are still limited to investigate a single level conditional dependency structure when there exists the multilevel structure among variables. Due to the fact that higher level variables may work together to accomplish certain tasks, simultaneously exploring conditional dependency structures among higher level variables and among lower level variables are of our main interest. Given multilevel data from heterogeneous classes, our method assures that common structures in terms of the multilevel conditional dependency are shared during the estimation procedure, yet unique structures for each class are retained as well. Our proposed approach is achieved by first introducing a higher level variable factor within a class, and then common factors across classes. The performance of our approach is evaluated on several simulated networks. We also demonstrate the advantage of our approach using breast cancer patient data. / Ph. D.

Submitted in fulfillment for the degree of Master of Technology, Clinical Technology: Cardiovascular Perfusion, Durban University of Technology, Durban, South Africa, 2016. / Introduction: A large volume of residual haemodilute blood remains in the cardiopulmonary bypass (CPB) circuit after termination of the bypass. It is common practice in many centres to process residual pump blood with an autologus cell salvage system (ACSS), thereby producing a re-suspended red blood cell (RBC) concentrate and attenuating the need for donor blood RBC concentrate. It has also become standard practice to wash donor pack red blood cells (PRBC) before adding it to neonate cardiopulmonary circuits (Swindell et al., 2007). Manufactures of ACSS recommend 0.9% sodium chloride (NaCl) as a wash solution for processing salvaged blood. Previous studies have demonstrated that washing PRBC with normal saline results in acid-base (Huber et al., 2013) and electrolyte derangements (Varghese et al., 2007). Infusion of normal saline in healthy volunteers also results in significant changes in osmolality (Williams et al., 1999). The use of normal saline as a wash solution in processing residual CPB blood requires investigation. Aims and Objectives: This was a prospective, quantitative in vitro investigation to analyze and compare the quality of residual pump blood post CPB that had been washed with either an unbalanced electrolyte solution (0.9% normal saline) or a balanced electrolyte solution (Balsol®). Both are crystalloid solutions. The primary objective of the present study was to measure and compare the pH, electrolytes, metabolites, osmolality and strong ion difference (SID) of residual pump blood to the pH, electrolytes, metabolites, osmolality and SID of processed cell saver blood, which was washed with either 0.9% normal saline or Balsol® solution. The secondary objective was to measure and compare protein levels (albumin and total protein) in residual pump blood to protein levels in processed cell saver blood, that is washed with either 0.9% normal saline or Balsol® solution. The final objective was to determine the volume, haematocrit and haemoglobin yield post cell saver processing, from the input volume of residual pump blood when washed with either 0.9% normal saline or Balsol® solution. This was the first study of this nature done in the South African population group. Methodology: In this investigation in a series of forty patients (n=40) undergoing elective cardiac surgery with CPB, the first twenty patients were allocated to the NaCl control group (n=20) and the second twenty patients were allocated to the Balsol® interventional group (n=20). The extracorporeal circuit consisted of a standard integral hollow fibre membrane oxygenator and tubing that was primed with 1500-1800 millilitres of balanced crystalloid solution (Balsol®), for both the control group and the interventional group, and addition of 5000 iu heparin. The balanced crystalloid solution (Balsol®) is the approved standard CPB priming solution for all cardiac procedures at Inkosi Albert Luthuli Central Hospital. This setup was used with the Stockert S5 roller pump heart lung machine. The operations were performed as per protocol with standard non-pulsatile CPB and hypothermia was maintained at 28 – 32 ºC (core) and haemodilution (haematocrit 20 % to 30 %). A standard flow rate of 2.4 L/min/m² was used. Cardio protection consisted of either cold Blood Cardioplegia using the Buckberg 4:1 ratio, being four parts blood to one part cardioplegia (with the 35ml of 20 % Dextrose + 1 gram Magnesium Sulphate added per 500ml), or 20ml/kg cold St Thomas II cardioplegia (with addition of 10ml of 8.5% NaHCO3 + 100mg lignocain per litre). Topical cooling was achieved with ice cold 0.9 % saline. Maintenance fluid used during CPB was Balsol® for both the control and the interventional groups. Calcium, potassium and sodium bicarbonate was administered as required during CPB to correct deficits for both groups. Weaning of CPB was performed after re-warming to a rectal temperature of at least 35 ºC for both study groups. Immediately on termination of CPB a blood sample was taken from the sampling manifold of the CPB circuit for pre wash analysis. Residual pump blood was then flushed out with one litre of Balsol® solution for both groups and collected into the Medtronic autolog cell saver reservoir to be processed. In the control study group 0.9% NaCl was used as the wash solution and in the interventional study group Balsol® solution was used as the wash solution. After processing of the salvaged blood is complete, a blood sample was taken for post wash analysis. Clinical data recorded for pre and post wash samples included: pH, pCO2, pO2, [K+], [Na+], [Cl-], [Ca2+], lactate, glucose, [HCO3-], TCO2, haematocrit, haemoglobin (GEM 4000® premier™ blood gas analyser) blood volume (Medtronic autolog) and SID (calculated as per equation). Inorganic phosphate, total magnesium, albumin, total protein (Siemens Advia 1800 blood gas analyser) and osmolality (Gonotech osmometer) were also measured. Results: There was a highly significant decrease (p < 0.05) within the NaCl group after washing with pCO2 (28.3 ± 2.9 vs. <6.0 ± 0.0), [K+] (4.5 ± 0.5 vs. 1.0 ± 0.7), total magnesium (1.7 ± 0.7 vs. 0.29 ± 0), ionized calcium (1.0 ± 0.09 vs. 0.1 ± 0.03), inorganic phosphate (0.9 ± 0.4 vs. 0.09 ± 0.04) and SID (27.1 ± 2.1 vs. 18.4 ± 2.2). There was a highly significant increase (p < 0.05) within the NaCl group after washing with pH (7.5 ± 0.1 vs. 7.7 ± 0.1), [Na+] (132.9 ± 3.2 vs. 146.3 ± 1.9), [Cl-] (107.8 ± 3.1 vs. 127.4 ± 2.1) and osmolaltity (256.9 ± 38.4 vs. 296.2 ± 57.5). There were highly significant decrease (p < 0.05) within the Balsol® group after washing with pCO2 (30.15 ± 6.0 vs. 18.9 ± 4.9), [Na+] (134.7 ± 2.2 vs. 125.6 ± 1), [Cl-] (108.8 ± 2.7 vs. 100.2 ± 1.4), ionized calcium (0.9 ± 0.1 vs. 0.02 ± 0.04), inorganic phosphate (0.8 ± 0.2 vs. 0.1 ± 0.024) and osmolality (288.8 ± 20.6 vs. 272.8 ± 19.9). There were highly significant increase (p < 0.05) within the Balsol® group after washing with pH (7.5 ± 0.1 vs. 7.7 ± 0.1), [K+] (4.2 ± 0.4 vs 4.6 ± 0.3). Total magnesium and SID were similar after washing within the Balsol® group. Albumin and total protein revealed similar significant decreases within both groups after washing. There was a highly significant difference (p < 0.05) in the change between groups after washing in all the variables measured, except for pH, inorganic phosphate, lactate, glucose, albumin, total protein, haematocrit, haemoglobin, and blood volume. Total carbon dioxide and [HCO3-] were not compared because they were incalculable by blood gas analyser in the NaCl group. Conclusion: This investigation concluded that the balanced electrolyte solution Balsol® used for washing residual CPB blood results in a re-suspended RBC concentrate, with an osmolality and electrolyte profile that is superior compared to washing residual CPB blood with 0.9% NaCl solution. / M

This dissertation thesis deals with the computational modelling of the dynamic response of the rotating electrical machine structure on the application of the magnetic forces. Apart from the dynamic response of the ideal symmetrical machine, the influence of the air gap eccentricity on the dynamics response is studied in this work. A basic type of the air gap eccentricity, which is caused by eccentric mounting of the rotor pack on the shaft of the rotor, is considered. The calculations the dependence of the magnetic forces on the time and a misalignment of the rotor pack are performed as first. The computational model of the magnetic field of the rotating electrical machine, which is based on solution of the electromagnetic coupled field analysis by finite element method, is used for this purpose. An analysis of the influence of the unbalanced magnetic pull and the stiffness of some parts of the machine on the modal properties of the machine is performed in the second part of this thesis. A third part of this thesis is focused on the calculation of the dynamic response of the machine during the steady state operation of the machine and the influence of the rotor pack misalignment on the dynamic response is studied. The obtained results showed that the tangential components of the magnetic forces, which act on the stator pack, excite significant torsional vibration of the stator. Besides the vibration of the stator of the machine, the influence of the rotor pack misalignment on the sound power of the machine, vibration of the rotor, loads of rotor bearings and air gap eccentricity is studied in this thesis.

No description available.

The simplified models of synchronous generators, widely used in stability studies of large electric power systems, are not completely suitable for the stability analysis and the design of controllers of distributed synchronous generators, generally connected to typically unbalanced branches. To more accurately analyze the systems with distributed generation, it is necessary to utilize synchronous generator models that consider frequency variation in their electrical equations. Furthermore, this model must represent possible unbalanced three-phase voltages at the generator terminals as well. Nonetheless, to provide reliable responses, the parameters of this more detailed model should be known. Thus, this work assesses the influence of the parameters on the responses of a detailed synchronous generator model, suitable to depict unbalanced operating conditions, and proposes an approach for the estimation of its most important parameters. In the proposed structure, we first employ Trajectory Sensitivity Functions to evaluate the dependency of the responses of this model with respect to its parameters and, from that, we rank them according to their importance. Subsequently, we apply an estimation process that utilizes the Unscented Kalman Filter with the aid of a genetic algorithm to estimate the main parameters of this synchronous generator model under unbalanced operating conditions. To obtain the results and, therefore, assess the proposed estimation approach, we make use of a system which comprises a synchronous generator connected to a three-phase unbalanced load. In addition to the unbalanced operation of the test system, we also consider noises due to the constant load switching, typical of distribution systems. The estimations performed for three operating conditions of the generator were very satisfactory, which demonstrates the efficiency of the proposed approach to obtain adequate models for the description of synchronous generator operation under unbalanced operating conditions. / Os modelos simplificados de geradores síncronos, amplamente utilizados em estudos de estabilidade de grandes sistemas elétricos de potência, não são completamente adequados para a análise de estabilidade e projetos de controladores dos geradores síncronos distribuídos, geralmente conectados a sistemas tipicamente desequilibrados. Para que os sistemas com geração distribuída possam ser analisados mais fidedignamente, é necessária a utilização de um modelo de gerador síncrono que considere a variação de frequência em suas equações elétricas. Além disso, esse modelo também deve ser capaz de representar possíveis tensões trifásicas desequilibradas nos terminais do gerador. Entretanto, para que esse modelo mais detalhado possa fornecer respostas coerentes com a realidade, deve-se conhecer seus parâmetros. Dessa forma, este trabalho avalia a influência dos parâmetros nas respostas de um modelo de gerador síncrono mais detalhado, adequado para representar operações desbalanceadas, e propõe uma abordagem para a estimação de seus parâmetros mais importantes. Nessa estrutura, inicialmente empregam-se as Funções de Sensibilidade de Trajetória para avaliar a dependência das respostas desse modelo em relação aos seus parâmetros e, a partir disso, ordená-los conforme sua importância. Em seguida, aplica-se um processo de estimação que utiliza o Filtro de Kalman Unscented com o auxílio de um algoritmo genético para estimar os principais parâmetros desse modelo de gerador síncrono em condições de desbalanço. Para a obtenção dos resultados e consequente avaliação da abordagem de estimação proposta, utiliza-se um sistema composto por um gerador síncrono conectado a uma carga trifásica desbalanceada. Além da operação desbalanceada desse sistema teste, também são considerados ruídos devidos ao constante chaveamento de cargas, típicos de sistemas de distribuição. As estimações realizadas para três condições de operação do gerador foram bem satisfatórias, indicando a eficiência da abordagem proposta na obtenção de modelos adequados para descrever a operação de geradores síncronos em condições de desbalanço.