From a fresh data science perspective, this thesis discusses the prediction of coronary artery disease based on Single-Nucleotide Polymorphisms (SNPs) from the Ontario Heart Genomics Study (OHGS). First, the thesis explains the k-Nearest Neighbour (k-NN) and Random Forest learning algorithms, and includes a complete proof that k-NN is universally consistent in finite dimensional normed vector spaces. Second, the thesis introduces two dimensionality reduction techniques: Random Projections and a new method termed Mass Transportation Distance (MTD) Feature Selection. Then, this thesis compares the performance of Random Projections with k-NN against MTD Feature Selection and Random Forest for predicting artery disease. Results demonstrate that MTD Feature Selection with Random Forest is superior to Random Projections and k-NN. Random Forest is able to obtain an accuracy of 0.6660 and an area under the ROC curve of 0.8562 on the OHGS dataset, when 3335 SNPs are selected by MTD Feature Selection for classification. This area is considerably better than the previous high score of 0.608 obtained by Davies et al. in 2010 on the same dataset.

Representações visuais têm sido adotadas na exploração de conjuntos de documentos, auxiliando a extração de conhecimento sem que seja necessária a análise individual de milhares de textos. Mapas de documentos, em particular, apresentam documentos individualmente representados espalhados em um espaço visual, refletindo suas relações de similaridade ou conexões. A construção destes mapas de documentos inclui, entre outras tarefas, o posicionamento dos textos e a identificação automática de áreas temáticas. Um desafio é a visualização de conjuntos dinâmicos de documentos. Na visualização de informação, é comum que alterações no conjunto de dados tenham um forte impacto na organização do espaço visual, dificultando a manutenção, por parte do usuário, de um mapa mental que o auxilie na interpretação dos dados apresentados e no acompanhamento das mudanças sofridas pelo conjunto de dados. Esta tese introduz um algoritmo para a construção dinâmica de mapas de documentos, capaz de manter uma disposição coerente à medida que elementos são adicionados ou removidos. O processo, inerentemente incremental e de baixa complexidade, utiliza um espaço bidimensional dividido em células, análogo a um tabuleiro de xadrez. Resultados consistentes foram alcançados em comparação com técnicas não incrementais de projeção de dados multidimensionais, tendo sido a técnica aplicada também em outros domínios, além de conjuntos de documentos. A visualização resultante não está sujeita a problemas de oclusão. A identificação de áreas temáticas é alcançada com técnicas de extração de regras de associação representativas para a identificação automática de tópicos. A combinação da extração de tópicos com a projeção incremental de dados em um processo integrado de mineração visual de textos compõe um espaço visual em que tópicos e áreas de interesse são destacados e atualizados à medida que o conjunto de dados é modificado / Visual representations are often adopted to explore document collections, assisting in knowledge extraction, and avoiding the thorough analysis of thousands of documents. Document maps present individual documents in visual spaces in such a way that their placement reflects similarity relations or connections between them. Building these maps requires, among other tasks, placing each document and identifying interesting areas or subsets. A current challenge is to visualize dynamic data sets. In Information Visualization, adding and removing data elements can strongly impact the underlying visual space. That can prevent a user from preserving a mental map that could assist her/him on understanding the content of a growing collection of documents or tracking changes on the underlying data set. This thesis presents a novel algorithm to create dynamic document maps, capable of maintaining a coherent disposition of elements, even for completely renewed sets. The process is inherently incremental, has low complexity and places elements on a 2D grid, analogous to a chess board. Consistent results were obtained as compared to (non-incremental) multidimensional scaling solutions, even when applied to visualizing domains other than document collections. Moreover, the corresponding visualization is not susceptible to occlusion. To assist users in indentifying interesting subsets, a topic extraction technique based on association rule mining was also developed. Together, they create a visual space where topics and interesting subsets are highlighted and constantly updated as the data set changes

Ferramentas e técnicas de visualização promovem uma análise de dados mais efetiva pelo fato de explorar a capacidade humana na percepção de padrões, principalmente em representações gráficas. Muitos fenômenos são associados a algum tipo de referência, temporal ou geográfica, que pode oferecer informação importante quando são submetidos a processos de análise. Este trabalho aborda representações visuais de dados geradas por técnicas de projeção multidimensional, e propõe uma estratégia para o tratamento diferenciado das referências temporais ou geográficas presentes em conjuntos de dados, no processo de gerar uma projeção multidimensional. Foi proposta e implementada uma variação da técnica Least Square Projection (LSP) que evidencia a informação das referências e permite ao usuário interagir com os mapas visuais gerados, bem como diversas funcionalidades que auxiliam no processo de análise exploratória. A nova abordagem é ilustrada por meio de estudos de caso envolvendo bases de dados temporais e com referências geográficas, em que foi possível observar o comportamento global dos elementos, bem como comportamentos de elementos ou grupos de elementos de interesse. Limitações da estratégia proposta também são discutidas / Visualization tools and techniques promote more effective data analysis by exploiting the human visual perception capabilities in detecting patterns in graphical representations. Many phenomena generate data that include temporal or geographical references, which are likely to provide important information in data analysis procedures. This work addresses data visualizations generated with multidimensional projections, proposing a strategy to handle temporal and geographical references present in multidimensional data sets, when generating multidimensional projections. The Least Squares Projection (LSP) technique was extended to explicitly handle the reference information and represent it in the visual maps, and a set of supporting analysis functions have been implemented. The proposed approach is illustrated through case studies on multidimensional data sets, in which it was possible to observe the global behavior of the elements, as well as individual behavior of elements or groups of elements of interest

Le Modèle Standard de la physique des particules décrit les interactions des particules sub-atomiques à travers les interactions fortes, faibles et électromagnétiques.Toutefois, il est clair que cette théorie ne constitue pas une description complète de la Nature. Mais le fait expérimental que les neutrinos changent de saveur au cours de leur propagation(oscillation des neutrinos) implique qu'ils ont une masse non nulle. Le travail de cette thèse se concentre sur l'expérience d'oscillation T2K : une expérience à longue ligne de base installée au Japon. % qui est détaillés dans le Chapitre 2. Pour la première fois T2K a pu observer l'apparition de neutrinos de saveur électronique dans le faisceau principalement composé de neutrinos muoniques etobtenir les premières contraintes sur la violation de la symétrie Charge-Parité. Il permet également de mesurer précisément les paramètres d'oscillations θ₁₃ , θ₂ ₃ et Δm²₂ ₃ .Cela nécessite une excellente compréhension du détecteur et des modèles d'interactions des neutrinos. Cette thèse se concentre sur ces deux objectifs : réduire les systématiques liées audétecteur et approfondir nos connaissances sur l'interaction des neutrinos avec la matière.Le groupe CEA, dans lequel je suis impliqué, est responsable de la maintenance et de l'opération des Chambres à Projections Temporelles (TPC) du détecteur proche (ND280).Elles permettent l'identification des particules chargées produites par l'interaction des neutrinos dans le detecteur et la mesure de leur impulsion.Une partie de mon travail a consisté en l'étude des TPCs et en particulier de l'alignement des modules MicroMegas installés sur les plans de lecture. En effet tout défaut d'alignement entre différents modules peut causer un biais sur la mesure del'impulsion des particules traversant la TPC. Les neutrinos sont étudiés grâce à l'observation de l'état final de leur interaction avec la matière. Les modèles d'interactions doivent alors être parfaitement bien compris,car les expériences d'oscillation présentes et futures approchent d'une phase où notre connnaissance des interactionsdes neutrinos devient un facteur limitant pour la détermination des paramètres d'oscillations. L'interaction quasi élastique par courant chargé (CCQE) est le processus dominant pour T2K. Ce manuscrit décrit la mesure durapport des sections efficaces d'interaction des neutrinos sur les noyaux de Carbone et d'Oxygène. Cette mesure contribue à la réduction des incertitudes pour l'analyse d'oscillation, liéesà l'utilisation d'une différente cible pour le détecteur proche et le détecteur lointain. / The Standard Model of particle physics describes the interactions of subatomic particles through the strong, weak and electromagneticinteractions. However, it is known that this theory is not a complete description of the nature. Indeed the observation that neutrinos can change their flavor alongtheir propagation path (neutrino oscillation) proves that they actually have a mass. The work of this thesis has been performed in the T2K oscillation experiment: a long baseline experiment located in Japan. % detailed in Chapter ref{sec_t2k_res}.Using the data collected so far, T2K has been able to observe for the first time the $nu_{mu} rightarrow nu_e$ appearance and to give first results on the Charge-Paritysimmetry in neutrino oscillation. It provides also precise measurements of the oscillation parameters θ₁₃ , θ₂ ₃, Δm²₂ ₃.The precise measurement of oscillation parameters requires a good understanding of the detector and of the neutrino interaction model. This thesis is thus focused on boththese aspects: reduce the detector uncertainties and improve our knowledge of neutrino interactions with matter.The CEA Saclay group where I am involved, is in charge of the maintenance and operation of the Time Projection Chambers (TPCs) of the T2K near detector (ND280)which are used to identify and measure the kinematics of the charged particles produced in neutrino interactions.Part of my work was focused on the study of the TPC and in particular on the alignment of the MicroMegas modules instrumenting their readout planes.Indeed, misalignments between modules may cause a bias on the momentum measurement of the particles crossing the TPC.Neutrinos are studied through the observation of the final state of their interactions with matter. Therefore, the interactionmodels need to be extremely well understood to infer the neutrino properties correctly.Indeed current and next future oscillation experiments are approaching the phase of precise measurements of the mixing parameters becominglimited by our knowledge of neutrino interactions.The charged current quasi-elastic (CCQE) interaction is the most relevant process at T2K.This manuscript describes the measurement of CCQE cross-section ratio between oxygen and carbon nuclei. It can contribute to reduce the uncertainties on the oscillation analysis arisingfrom the different target between the near and far detector.

MENVSC (Geography) / Department of Geography and Geo-Information Sciences / Rainfall is undoubtedly the most significant factor for life’s continuity. South Africa is prone to future climate uncertainties due to global climate change. The aim of this study is to investigate rainfall variability and change in South Africa on a present day (1976-2005), near-future (2006-2035) and far-future (2036-2065) climate. For the study, 3 RCMs (REMO2009, RCA4 and CCLM4-8-17), forming part of CORDEX-Africa project were nested within 5 different CIMP5_GCMs of low resolution. GPCC precipitation, NOAA GHCN_CAMS Land Temperature and other NCEP reanalysis products were useful in validating models in simulations of present-day climate. RCP4.5 and RCP8.5 emission scenarios from IPCC-AR5 were used for future climate projections. On the validation, each regional climate model displayed different signature on simulations, rainfall in particular because this is a variable that is affected most by sub-grid process. Simulations nested within MIROC5 simulated more precipitation than simulations forced with other GCMs, due to more large-scale moisture convergence into the nested domain. There were differences in projections of RCM nested within the same GCM, as well as with the same RCM nested within different GCMs, on the future. Models nested within MPI project wetter conditions over the eastern parts of Limpopo, while the other two projected drier conditions in the same area. REMO2009 forced on MPI uniquely projected drying of Western Cape throughout the seasons on both RCPs and futures. Simulations conducted with the RCP8.5 scenario forcing are generally found to be associated with either a larger increase in temperature, or an increase in area associated with higher temperature increases. CCLM4-8-17 forced on HadGEM2 projected below average temperatures over the northwest parts of the country under the RCP8.5 scenarios. MPI driving model projected a general reduction of evaporation values, with lowest over northeast, northwest parts and south coastal parts of South Africa, in contrary to adjacent oceans. In this study, we have sought to identify the sources of uncertainties amongst model simulations between either the RCMs or the driving GCMs. / NRF

Land-cover and land-use change are highly dynamic and contribute to changes in the water balance. The most common changes are urbanisation, deforestation and desertification. This dissertation deals with the topic of projecting land cover (LC) into the near future with the help of the scenario technique. The aim of the thesis is the projection of the urban and rural land-cover change (LCC) till 2025. Two research questions are addressed in this work: (1) Which integrated concept can be developed to combine different methods to project urban and rural LCC into the future based on past LCC? (2) Is it possible to implement the developed concept and does the implementation deliver plausible results? To answer the research questions, a 4-step concept is adopted which serves as workflow for projecting the LCC: (i) the definition of the scenario context, and with that the definition of the study area, (ii) the identification of spatial and dynamic drivers for LCC, consisting of spatial drivers that are location-dependent, such as slope or soil type, and dynamic drivers of LCC, such as demographic and economic development, (iii) scenario formulation and projection of identified drivers, and (iv) scenario-based projections of future LCC, which means its quantitative and spatial modification (demand and allocation). For implementation and testing, the Upper Western Bug River catchment in Ukraine serves as the study site. The extent of the study area reaches from the source of the Western Bug to the Dobrotvir gauging station and is thus entirely located in Ukraine. This presents the first step of the developed concept of the projection of LCC. The existing geo-database for implementation is scarce. LC data is available for the territory of the EU (e.g. CORINE Land Cover) but not for Ukraine. Therefore, the implementation of the second step had to focus on the derivation of LC data for three-time steps to get the basis for the LCC. A classification of satellite scenes of Landsat and SPOT are done for the time steps 1989, 2000, and 2010. The two decades show a huge development of LCC. The increase of ‘artificial surface’ and unmanaged ‘grassland’ is visible with the decrease of ‘arable land’ and ‘forests’. An extended statistical analysis considering the systematic LCC reveals stable transition pathways, which in turn are the basis of the projection of future land cover. This refers to the second step of the concept: change detection. One transition pathway is that ‘arable land’ is not used and converted in settlement areas, but rather changes into ‘grassland’. With the derived LC and the analysed LCC as a basis of the work, the search for spatial and dynamic drivers start at the third time step. A list of dynamic drivers is first compiled, via literature research, and then tested for effect on LCC with statistical analyses. The dynamic driving forces are the ‘Gross Domestic Product’ (GDP) and ‘population development’. Spatial driving forces are laws/planning practices, fertility, slope, distance to the city Lviv, settlements, roads, or rivers. As a result, population development has an effect on the change in the LC class to 'artificial surface' from 'grassland' and 'arable land' from ‘grassland'. The implementation of the third step is done with the help of four storylines where the overall development of the dynamic drivers are included towards 2025. With that it is possible to project them into the future. The fourth step includes the calculation of the demand for each LC class with the projected dynamic drivers. The areas that have a high probability to change into another LC class are determined in suitability maps (allocation) which are derived by translating the transition pathways into GIS algorithms including the spatial driving forces. The class of 'artificial surface' changes the most under scenario A until 2025 and less under scenario D — the sustainable scenario. The LC class 'arable land' decreases in scenario A and B, but has the strongest development in scenario D. The LC class of unmanaged ‘grassland’ is quite stable under scenario A and B, but decreases in C and D. The results of systematic changes in ‘arable land’ that changes into ‘grassland’ are different compared to developments in other countries like Germany. The protection and conservation of arable land is not seen as strongly in other Eastern European countries as it is in the Upper Western Bug River catchment. In turn, the identified spatial and dynamic drivers fit other studies in Eastern Europe. The applied concept of projecting LCC with these steps are highly flexible for implementation in other study sites. However, the volume of work can differ within the steps because of the available databases. In Ukraine the available LCC data was not detailed enough to carry out a future projection. So, a main part of the work is dedicated to the derivation of past LC for different time steps. The involvement of regional experts helped to gain detailed knowledge of processes of LCC. The advantage of the presented concept with the mixture of quantitative (e.g. satellite analyses, statistical analyses) and qualitative methods can overcome methodological knowledge gaps. In addition, the retrospective analyses, as starting points, for the projection of future LCC carves out the site-specific allocation of change.:Acknowledgements..............................................................................................................................III Abstract..................................................................................................................................................IV Zusammenfassung..............................................................................................................................VII Contents..................................................................................................................................................X Abbreviations......................................................................................................................................XIV 1.Introduction.................................................................................................................................1 1.1Background................................................................................................................................1 1.2Objectives and Research Questions.......................................................................................3 1.3Structure....................................................................................................................................3 2.Basics of the Work......................................................................................................................5 2.1Land Cover, Land Use and Land-cover Change....................................................................5 2.2Projection of Land-Cover Change...........................................................................................6 2.3Drivers of Land-Cover and Land-Use Change.....................................................................10 2.4Basics of Scenario Methods..................................................................................................12 3.Conceptual Framework............................................................................................................15 3.1Step1: Definition of the Scenario Context...........................................................................17 3.2Step 2: Identification of Spatial and Dynamic Drivers of Land-Cover Change...............18 3.3Step 3: Scenario Formulation and Projection of Identified Drivers.................................18 3.4Step 4: Scenario-based Projections of Future Land-Cover Change.................................19 4.Implementation and Testing of the Framework..................................................................20 4.1Step 1: Definition of the Scenario Context..........................................................................20 4.2Step 2: Identification of Spatial and Dynamic Drivers for Land-Cover Change..............24 4.3Step 3: Scenario Formulation and Projection of Drivers...................................................28 4.4Step 4: Scenario-based Projections of future Land-Cover Change..................................32 5.Discussion..................................................................................................................................36 5.1Discussion of the Methods....................................................................................................36 5.2Discussion of the Empirical Results.....................................................................................42 6.Conclusions and Outlook........................................................................................................47 7.Reference List............................................................................................................................49 8.Appendix....................................................................................................................................58 8.1Position and Affiliation of the Interviewed Experts...........................................................59 8.2Suitability Maps.......................................................................................................................60 8.3Research Articles.....................................................................................................................66 8.3.1Research Article 1: Retrospective Analysis of Systematic Land-Cover Change in the......... Upper Western Bug River catchment, Ukraine.....................................................................67 8.3.2Research article 2: Cross-Sectoral Projections of Future Land-Cover Change for the........ Upper Western Bug River catchment, Ukraine.....................................................................80

El presente modelo de negocio consiste en la importación de artículos o aparatos electrónicos y tecnológicos producidos en China para ser comercializados al por mayor a comerciantes minoristas en el mercado peruano, a través de un portal web, teniendo en cuenta los requerimientos y características que se necesitan, según la percepción de calidad que tiene el cliente. Para esto se elaboró un sondeo a distribuidores minoristas ubicados en centros comerciales en Lima Metropolitana, sobre calidad, variedad y distribución de productos tecnológicos entre otros; en la que podemos observar resultados importantes a tomar en cuenta: un 66% de encuestados, considera que la calidad de productos podría mejorar; un 64%, considera que no les ofrecen los productos necesarios para su negocio y un 47%, considera mala la distribución de productos. Para el abastecimiento, se ha previsto la selección de proveedores de China, mediante un plan de importación semestral. El plan de marketing se sintetiza en la diferenciación de nuestros productos, distribución y su servicio post venta, para ello se tiene previsto gastos de publicidad en marketing digital, contando con un portal web especializado para las ventas online. La inversión del proyecto totaliza casi S/ 300,000. La estructura financiera estaría conformada por el aporte de capital de 3 socios en partes iguales. Las proyecciones financieras efectuadas a valores constantes se contemplan con un horizonte de evaluación de 5 años. Los indicadores de rentabilidad revelan índices favorables para el proyecto con una TIR del 95.20%, un VAN de S/ S/ 2,091,072. / This business model consists of the importation of electronic and technological articles or devices produced in China to be sold wholesale in the Peruvian market, through a web portal, taking into account the requirements and characteristics that are needed, according to the perception of quality that the client has. For this, a survey of retail distributors located in shopping centers in Metropolitan Lima was carried out on quality, variety and distribution of technological products, among others; in which we can observe important results to take into account: 66% of respondents consider that the quality of products could improve; 64% consider that they are not offered the products necessary for their business and 47% consider the product distribution bad. For the supply, the selection of suppliers from China has been foreseen, with a biannual import plan. The Marketing Plan is synthesized in the differentiation of our products, distribution and their after-sales service, for this, advertising expenses in digital marketing are planned, with a specialized web portal for online sales. The project investment totals almost S / 300,000. The financial structure would be made up of the capital contribution of 3 partners in equal parts. Financial projections made at constant values ​​are contemplated with an evaluation horizon of 5 years. The profitability indicators reveal favorable indices for the project with an IRR of 44.52%, a NPV of S / 842,640. / Trabajo de investigación

The field of Deep Learning (DL) only over the last decade has proven useful for increasingly more complex Machine Learning tasks and data, a notable milestone being generative models achieving facial synthesis indistinguishable from real faces. With the increased complexity in DL architecture and training data, follows a steep increase in time and hardware resources required for the training task. These resources are easily accessible via cloud-based platforms if the data owner is willing to share its training data. To allow for cloud-sharing of its training data, The Swedish Transport Administration (TRV) is interested in evaluating resource effective, infrastructure independent, privacy-preserving obfuscation methods to be used on real-time collected data on distributed Internet-of-Things (IoT) devices. A fundamental problem in this setting is to balance the trade-off between privacy and DL utility of the obfuscated training data. We identify statistically measurable relevant metrics of privacy achievable via obfuscation and compare two prominent alternatives from the literature, optimization-based methods (OBM) and random projections (RP). OBM achieve privacy via direct optimization towards a metric, preserving utility-crucial patterns in the data, and is typically in addition evaluated in terms of a DL-based adversary’s sensitive feature estimation error. RP project data via a random matrix to lower dimensions to preserve sample pair-wise distances while offering privacy in terms of difficulty in data recovery. The goals of the project centered around evaluating RP on privacy metric results previously attained for OBM, compare adversarial feature estimation error in OBM and RP, as well as to address the possibly infeasible learning task of using composite multi-device datasets generated using independent projection matrices. The last goal is relevant to TRV in that multiple devices are likely to contribute to the same composite dataset. Our results complement previous research in that they indicate that both privacy and utility guarantees in a distributed setting, vary depending on data type and learning task. These results favor OBM that theoretically should offer more robust guarantees. Our results and conclusions would encourage further experimentation with RP in a distributed setting to better understand the influence of data type and learning task on privacy-utility, target-distributed data sources being a promising starting point. / Forskningsområdet Deep Learning (DL) bara under det senaste decenniet har visat sig vara användbart för allt mer komplexa maskinginlärnings-uppgifter och data, en anmärkningsvärd milstolpe är generativa modeller som erhåller verklighetstrogna syntetiska ansiktsbilder. Med den ökade komplexiteten i DL -arkitektur och träningsdata följer ett kraftigt ökat behov av tid och hårdvaruresurser för träningsuppgiften. Dessa resurser är lättillgängliga via molnbaserade plattformar om dataägaren är villig att dela sin träningsdata. För att möjliggöra molndelning av träningsdata är Trafikverket (TRV) intresserat av att utvärdera resurseffektiva, infrastrukturoberoende, privatiserade obfuskeringsmetoder som ska användas på data hämtad i realtid via distribuerade Internet-of-Things ( IoT) -enheter; det grundläggande problemet är avvägningen mellan privatisering och användbarhet av datan i DL-syfte. Vi identifierar statistiskt mätbara relevanta mått av privatisering som kan uppnås via obfuskering och jämför två framstående alternativ från litteraturen, optimeringsbaserade metoder (OBM) och slumpmässiga projektioner (RP). OBM uppnår privatisering via matematisk optimering av ett mått av data-privatisering, vilket bevarar övriga nödvändiga mönster i data för DL-uppgiften. OBM-metoder utvärderas vanligtvis i termer av en DL-baserad motståndares uppskattningsfel av känsliga attribut i datan. RP obfuskerar data via en slumpmässig projektion till lägre dimensioner för att bevara avstånd mellan datapunkter samtidigt som de erbjuder privatisering genom teoretisk svårighet i dataåterställning. Målen för examensarbetet centrerades kring utvärdering av RP på privatiserings-mått som tidigare uppnåtts för OBM, att jämföra DL-baserade motståndares uppskattningsfel på data från OBM och RP, samt att ta itu med den befarat omöjliga inlärningsuppgiften att använda sammansatta dataset från flera IoT-enheter som använder oberoende projektionsmatriser. Sistnämnda målet är relevant i en miljö sådan som TRVs, där flera IoT-enheter oberoende bidrar till ett och samma dataset och DL-uppgift. Våra resultat kompletterar tidigare forskning genom att de indikerar att både privatisering och användbarhetsgarantier i en distribuerad miljö varierar beroende på datatyp och inlärningsuppgift. Dessa resultat gynnar OBM som teoretiskt sett bör erbjuda mer robusta garantier vad gäller användbarhet. Våra resultat och slutsatser uppmuntrar framtida experiment med RP i en distribuerad miljö för att bättre förstå inverkan av datatyp och inlärningsuppgift på graden av privatisering, datakällor distribuerade baserat på klassificerings-target är en lovande utgångspunkt.

No description available.

No description available.