Magister Scientiae (Biodiversity and Conservation Biology) - MSc (Biodiv and Cons Biol) / Small ruminants found in pastoral dryland systems face many challenges regarding constant environmental and climate changes. These variable environments are home to many pastoralists who use the natural resources to sustain their livestock. Due to the prevailing environmental changes, these pastoralists require various adaptation strategies for the survival of their small ruminants. Adaptation is vital for all pastoralists globally, especially those living in semi-arid to arid regions as these areas are badly impacted by climate change. Farming with small ruminants that are able to constantly adapt to variable environments are a long-term and low cost resource strategy for farmers in these areas. Drought is a recurring challenge that farmers in dryland systems have to cope with and small ruminants are forced to adapt to food and water available to them as drought affects their survival, reproduction and production. This is because drought affects the environment directly as rainfall has an impact on all forage and water resources found in drylands. Thus available forage and the utilisation of this resource by small ruminants is essential for sheep and goat farmers who depend on these animals for their livelihoods. This thesis sets out to establish sheep and goat breeds found in a dryland pastoral system in South Africa focussing on their genetic differences and their diet selection during a drought period. The study was carried out in the 582 634 ha Steinkopf communal area located in the semi-arid to arid region of Namaqualand in South Africa. The dominant land use in this area is livestock farming where farmers utilise two biomes for grazing. Using mitochondrial DNA, genetic differences were investigated in small ruminants found in Steinkopf. The small ruminants that were studied were the Boer goats, Swakara sheep (Karakul) and cross-bred sheep. These were the most commonly farmed small ruminants in this area. Genetic differences and diversity found within the small ruminants in this area supports their ability to adapt to drought and changing environments. Diet selection of these small ruminants were assessed using direct forage observation during this drought period in both the Succulent and Nama Karoo biomes. Findings of the study indicate that small ruminants during a drought period will largely forage on what is available to them. Boer goats changed their diets from browser to grazers in the different biomes while sheep remained grazers in both biomes. This suggests that these animals are well adapted to changing forage availability and will adjust their feeding behaviours accordingly. Succulent plant species were utilised by all small ruminants and this can be regarded as new knowledge for small ruminants in dryland systems. Succulent species play a role in the diets of small ruminants in the Steinkopf rangeland and these animals have adapted traits that allow them to eat plants that contain high levels of phenolics and tannins, which show their high adaptability to this area. This study also revealed that indigenous and locally bred sheep in this area are crucial resources for livestock farming in this area. Swakara sheep in this area showed a low level of genetic variation while cross-bred sheep and the Boer goat had significant variation. With a long tradition of breeding small ruminants in the Steinkopf communal area that suit changing conditions, livestock keepers have accumulated detailed knowledge of their animals. Recommendations from this study were to do further studies regarding genetic variation within the cross-bred sheep as this information will be useful for breeding programs in dryland areas and it will add to the small ruminant genetics in South Africa. Because this study was conducted in a drought period, I recommend that diet selection should be observed during a regular wet and dry period as this would indicate how these animals adapt to what forage is available to them. These studies can act as management strategies for herders as it can provide information on how to use the natural resources sustainably.

Research report to School of Electrical and Information Engineering / An investigation was conducted into how a variable speed drive can provide a damping torque when mechanical oscillations are present. The modeling of mechanical oscillations via an analogous electrical circuit was performed. Simulation was used to demonstrate how a variable speed drive is able to damp speed oscillations using Direct Torque Control (DTC). Damping of mechanical oscillations is done by means of the variable speed drive providing a damping torque component that is in-phase with the speed deviation. The simulation showed that by applying a small torque component with the speed variation results in torque oscillations being damped by 60% after the initial disturbance. Damping is further improved by applying a torque component equal to the speed variation resulting in the oscillations being damped by 80% when compared to the initial disturbance. / MT2017

Cette thèse s’articule autour de trois essais. Le premier chapitre analyse l'impact des 27 ans de guerre civile en Angola sur les dépenses par équivalent adulte, le capital humain et la fécondité. La prédiction des effets de la guerre se fait à l’aide d’un modèle néoclassique de ménage unitaire dans la tradition de Rosenzweig. A partir de l’approche d’estimation par variable instrumentale, cette thèse montre que la guerre civile a un impact négatif et désastreux à court terme sur la santé des enfants, cet effet est persistant. La guerre civile n’a pas d’impact sur les dépenses par équivalent adulte. Elle accroit lascolarisation et décroit la fécondité à court terme. Le second chapitre de cette thèse analyse l’efficacité d’un programme social dans un environnement en conflit comme celui de l’Angola et s’interroge si cette efficacité dépend de l’intensité du conflit. Notre stratégie d’identification est basée sur la géographie politique du déploiement du programme basée sur un modèle de compétition spatiale à la Hotelling. Cette thèse montre que le Fond Social Angolais a eu un impact positif sur les dépenses par équivalent adulte et sur l’une des principales mesures anthropométriques à savoir le z-Score de la taille pour âge. L’efficacité du programme en fonction de l’intensité du conflit est analysée à l’aide de l’estimateur de variable instrumental local. La thèse montre que l’efficacité du programme augmente avec l’intensité du conflit. Le dernier chapitre de cette thèse analyse dans le cas du Cameroun, l’impact de la formation des enseignants en matière de VIH/SIDA. Les deux critères retenus pour le choix des écoles participant au programme, nous amène à choisir comme stratégie d’identification la régression discontinue. Cette thèse montre que les filles âgées de 15 à 17 ans dans les écoles traitées sont moins susceptibles d’avoir une grossesse involontaire. Pour les élèves âgés de 12 à 13 ans, la probabilité d'abstinence auto déclarée et l'utilisation du préservatif est également significativement plus élevé dans les écoles traitées / This thesis is based on three essays. The first chapter analyses the impact of 27 years of civil war in Angola on human capital, expenditures per adult equivalent and fertility. The prediction of the effects of civil war is done through a neoclassical unitary household model in the tradition of Rosenzweig. Using instrumental variable method, this thesis shows that civil war has a negative and disastrous impact in short-Term on health of children, this effect is persistent. Civil war has no impact on expenditures per adult equivalent. It increases enrollment and decreases fertility in the short term. The second chapter ofthis thesis analyzes the effectiveness of a social program in a conflict country such as Angola and explores whether this effectiveness depends on the intensity of the conflict. Our identification strategy is based on the political geography of the deployment of the program based on a model of spatial competition of Hotelling. This thesis shows that the Angola Social Fund had a positive impact on expenditures per adult equivalent and on one of the main anthropometric measurements namely the height for age z-Score. The program's effectiveness in function to the intensity of the conflict is analyzed using the local instrumental variable estimator. The thesis shows that the program's effectiveness increases with the intensity of the conflict. The last chapter of this thesis analyzes in the case of Cameroon, the impact of teacher training on HIV/AIDS. The two criteria for selecting participating schools, leads us to choose as identification strategy the regression discontinuity design. This thesis shows that 15 to 17 year old girls in teacher training schools are between 7 and 10 percentage points less likely to have started childbearing. For 12 to 13 year old girls, the likelihood of self-Reported abstinence and condom use is also significantly higher in treated schools.

Este trabalho tem como objetivo a comparação dos resultados de ensaios de durabilidade em um suporte do eixo traseiro de um veículo leve de passeio, submetido a dois tipos distintos de testes: durabilidade em rodagem real e em simuladores de estradas. Após o término dos ensaios, o componente apresentou resultados diferentes: as trincas podem ser observadas na peça submetida à rodagem real, e os danos não ocorrem no eixo que foi testado em simulador. As discrepâncias citadas podem estar ligadas a possíveis erros, ocorridos durante a transferência dos dados pertinentes ao teste real para o simulador. / The aim of this research is comparing results from durability tests performed at a rear axle bracket of a small passenger car. The part was tested using two different kinds of tests: real (proving ground) durability and simulated (road simulator) durability. After that, the part showed different final results: several cracks can be observed at the part that was tested at the proving ground and no real damage was inflicted at the part that was test at the road simulator. These differences can be related to transferability problems that occurred during data analysis and transfer from real test to a simulated one.

The HydroBone is a variable stiffness load-bearing element, which utilizes jamming of granular media to achieve stiffness modulation, controlled by the application of positive pressure. Several compressive tests were conducted on the HydroBone in order to quantify the load-bearing capability of the system. It was determined that the stiffness of the HydroBone was a function of the internal pressure of the system. A controller was modeled based on this function to achieve automatic stiffness modulation of the HydroBone. An exoskeleton was designed based on the HydroBone and various actuators for the exoskeleton were considered. The HydroMuscle, a soft linear actuator was selected to provide knee actuation for the exoskeleton, based on several efficiency and force output test conducted. A knee brace was designed, capable of producing 15Nm of torque on the knee, actuated using Bowden cables coupled to the HydroMuscles.

In a wide range of applications, datasets are generated for which the number of variables p exceeds the sample size n. Penalised likelihood methods are widely used to tackle regression problems in these high-dimensional settings. In this thesis, we carry out an extensive empirical comparison of the performance of popular penalised regression methods in high-dimensional settings and propose new methodology that uses ensemble learning to enhance the performance of these methods. The relative efficacy of different penalised regression methods in finite-sample settings remains incompletely understood. Through a large-scale simulation study, consisting of more than 1,800 data-generating scenarios, we systematically consider the influence of various factors (for example, sample size and sparsity) on method performance. We focus on three related goals --- prediction, variable selection and variable ranking --- and consider six widely used methods. The results are supported by a semi-synthetic data example. Our empirical results complement existing theory and provide a resource to compare performance across a range of settings and metrics. We then propose a new ensemble learning approach for improving the performance of penalised regression methods, called STructural RANDomised Selection (STRANDS). The approach, that builds and improves upon the Random Lasso method, consists of two steps. In both steps, we reduce dimensionality by repeated subsampling of variables. We apply a penalised regression method to each subsampled dataset and average the results. In the first step, subsampling is informed by variable correlation structure, and in the second step, by variable importance measures from the first step. STRANDS can be used with any sparse penalised regression approach as the ``base learner''. In simulations, we show that STRANDS typically improves upon its base learner, and demonstrate that taking account of the correlation structure in the first step can help to improve the efficiency with which the model space may be explored. We propose another ensemble learning method to improve the prediction performance of Ridge Regression in sparse settings. Specifically, we combine Bayesian Ridge Regression with a probabilistic forward selection procedure, where inclusion of a variable at each stage is probabilistically determined by a Bayes factor. We compare the prediction performance of the proposed method to penalised regression methods using simulated data.

Orientador: Daniela Renata Cantane / Resumo: Originadas na década de 1970, as Infecções Hospitalares vêm cada vez mais tomando proporções colossais, acarretando óbito em cerca de 30% dos pacientes em Unidades de Terapia Intensiva (UTI). Os pacientes diagnosticados com a infecções permanecem muito tempo internados, gerando um custo muito alto para os hospitais. No ambiente hospitalar a bactéria Acinetobacter baumannii a principal responsável por tais infecções, devido a sua facilidade de sobreviver em ambientes secos e úmidos, podendo sobreviver tanto no organismo humano, quanto nos ambientes que os profissionais da saúde entram em contato (computadores, equipamentos médicos, etc). Os principais vetores desta bactéria são os próprios agentes de saúde, visto que os pacientes na UTI estão todos acamados. No entanto, medidas de higienização extremamente necessárias para conter surtos da infecções o. Por outro lado, devido as emergências nestas unidades, muitas vezes não há tempo hábil para tais procedimentos. Visto que impossível uma medida total de higienização e uma taxa nula de contato da equipe de trabalho com o ambiente em UTIs, importante conhecer quais são as mínimas medidas necessárias para a diminuição de infecções hospitalares. Neste contexto, o objetivo deste trabalho propor e analisar um modelo que descreva a dinâmica de transmissão da infecção dentro de uma UTI, considerando pacientes e profissionais da saúde, assim como, propor um modelo de otimização visando determinar quais as mínimas medidas de higienização... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Originated in the 1970s, Hospital Infections come every time more taking colossal proportions, causing death in about 30% of patients in Intensive Care Units (ICU). Patients diagnosed with infections remain long hospitalized, generating a very high cost for hospitals. In the hospital environment, Acinetobacter baumannii is the main responsible for such infections due to their ease of survival in dry and humid, and can survive both in the human body and in the environments that health workers contact (computers, medical equipment, etc.). The main vectors of this bacterium are the health agents themselves, since the patients in the ICU are all bedridden. However, hygiene measures are extremely necessary to contain outbreaks of infection. On the other hand, due to emergencies in these units, there is often no time for such procedures. Since a total sanitation measure and it is important to know the minimum measures necessary for the decrease of infections. In this context, the objective xiii of this work is to propose and analyze a model that describes the dynamics of transmission of infection within an ICU, considering patients and health professionals, as well as to propose an optimization model aiming to determine the minimum hygienic measures are needed to minimize the number of infected patients. A Variable Neighborhood Search Metaheuristic was proposed to solve the optimization model. For validation of the models were carried out computational simulations. These simulation... (Complete abstract click electronic access below) / Mestre

Thesis. 1976. M.S.--Massachusetts Institute of Technology. Dept. of Ocean Engineering. / Microfiche copy available in Archives and Engineering. / Includes bibliographical references. / M.S.

Soft robots present the opportunity to extend the capabilities currently demonstrated within the field of robotics. By utilizing primarily soft materials in their construction, soft robots are inherently safe to operate around humans, can handle delicate tasks without advanced controls, and are robust to shocks and impacts during deployment. While proof-of-concept devices have been demonstrated successfully, there remains a need for widely applicable, reliable soft robotic components. This dissertation presents sensors to reliably measure the large deformations exhibited in soft robotic structures and responsive structures enabled by variable stiffness materials that can switch from flexible to stiff on-demand. By characterizing the sensors from the material level, through the manufacturing, to the completed functional device, the fabrication processes can be depended upon to produce sensors with predictable, reliable performance. The sensors were applied to various soft robotic systems through implementation on the surface of the structures to measure surface strains, and embedded in the body of the robot to measure body deformations. The sensory feedback was used to reconstruct the state of and to perform closed-loop control of the soft robot's position. Variable stiffness materials that switch from rigid to soft through application of heat were leveraged to create responsive structures that can be deformed or reconfigured on-demand. This capability is necessary for soft robots to exert load onto the external environment and enables a wider range of interactions with target objects. The work presented in this dissertation furthers the field of soft robotics by illustrating a path toward proven, reliable soft sensors for measuring large strains and variable stiffness materials to create responsive structures.

L'objectif de cette thèse est de proposer un ensemble d'outils méthodologiques pour répondre à deux problématiques : la prédiction de masse salariale des collectivités, et l'analyse de leurs données de fiscalité. Pour la première, les travaux s'articulent à nouveau autour de deux thèmes statistiques : la sélection de modèle de série temporelle, et l'analyse de données fonctionnelles. Du fait de la complexité des données et des fortes contraintes de temps de calcul, un rassemblement de l'information a été privilégié. Nous avons utilisé en particulier l'Analyse en Composantes Principales Fonctionnelle et un modèle de mélanges gaussiens pour faire de la classification non-supervisée des profils de rémunération. Ces méthodes ont été appliquées dans deux prototypes d'outils qui représentent l'une des réalisations de cette thèse. Pour la seconde problématique, le travail a été effectué en trois temps : d'abord, des méthodes novatrices de classification d'une variable cible ordinale ont été comparées sur des données publiques déjà analysées dans la littérature, notamment en exploitant des forêts aléatoires, des SVM et du gradient boosting. Ensuite, ces méthodes ont été adaptées à la détection d'anomalies dans un contexte ciblé, ordinal, non supervisé et non paramétrique, et leur efficacité a été principalement comparée sur des jeux de données synthétiques. C'est notre forêt aléatoire ordinale par séparation de classes qui semble présenter le meilleur résultat. Enfin, cette méthode a été appliquée sur des données réelles de bases fiscales, où les soucis de taille et de complexité des données sont plus importants. Destinée aux directions des collectivités territoriales, cette nouvelle approche de l'examen de leur base de données constitue le second aboutissement de ces travaux de thèse. / The aim of this thesis is to provide a set of methodological tools to answer two problems: the prediction of the payroll of local authorities, and the analysis of their tax data. For the first, the work revolves around two statistical themes: the selection of time series model, and the analysis of functional data. Because of the complexity of the data and the heavy computation time constraints, a clustering approach has been favored. In particular, we used Functional Principal Component Analysis and a model of Gaussian mixtures to achieve unsupervised classification. These methods have been applied in two prototypes of tools that represent one of the achievements of this thesis. For the second problem, the work was done in three stages: first, innovative methods for classifying an ordinal target variable were compared on public data, notably by exploiting random forests, SVM and gradient boosting. Then, these methods were adapted to outlier detection in a targeted, ordinal, unsupervised and non-parametric context, and their efficiency was mainly compared on synthetic datasets. It is our ordinal random forest by class separation that seems to have the best result. Finally, this method has been applied to real data of tax bases, where the concerns of size and complexity are more important. Aimed at local authorities directorates, this new approach to examining their database is the second outcome of this work.