Plusieurs applications modernes nécessitent un traitement de flux massifs de données XML, cela crée de défis techniques. Parmi ces derniers, il y a la conception et la mise en ouvre d'outils pour optimiser le traitement des requêtes XPath et fournir une estimation précise des coûts de ces requêtes traitées sur un flux massif de données XML. Dans cette thèse, nous proposons un nouveau modèle de prédiction de performance qui estime a priori le coût (en termes d'espace utilisé et de temps écoulé) pour les requêtes structurelles de Forward XPath. Ce faisant, nous réalisons une étude expérimentale pour confirmer la relation linéaire entre le traitement de flux, et les ressources d'accès aux données. Par conséquent, nous présentons un modèle mathématique (fonctions de régression linéaire) pour prévoir le coût d'une requête XPath donnée. En outre, nous présentons une technique nouvelle d'estimation de sélectivité. Elle se compose de deux éléments. Le premier est le résumé path tree: une présentation concise et précise de la structure d'un document XML. Le second est l'algorithme d'estimation de sélectivité: un algorithme efficace de flux pour traverser le synopsis path tree pour estimer les valeurs des paramètres de coût. Ces paramètres sont utilisés par le modèle mathématique pour déterminer le coût d'une requête XPath donnée. Nous comparons les performances de notre modèle avec les approches existantes. De plus, nous présentons un cas d'utilisation d'un système en ligne appelé "online stream-querying system". Le système utilise notre modèle de prédiction de performance pour estimer le coût (en termes de temps / mémoire) d'une requête XPath donnée. En outre, il fournit une réponse précise à l'auteur de la requête. Ce cas d'utilisation illustre les avantages pratiques de gestion de performance avec nos techniques / Many modern applications require processing of massive streams of XML data, creating difficult technical challenges. Among these, there is the design and implementation of applications to optimize the processing of XPath queries and to provide an accurate cost estimation for these queries processed on a massive steam of XML data. In this thesis, we propose a novel performance prediction model which a priori estimates the cost (in terms of space used and time spent) for any structural query belonging to Forward XPath. In doing so, we perform an experimental study to confirm the linear relationship between stream-processing and data-access resources. Therefore, we introduce a mathematical model (linear regression functions) to predict the cost for a given XPath query. Moreover, we introduce a new selectivity estimation technique. It consists of two elements. The first one is the path tree structure synopsis: a concise, accurate, and convenient summary of the structure of an XML document. The second one is the selectivity estimation algorithm: an efficient stream-querying algorithm to traverse the path tree synopsis for estimating the values of cost-parameters. Those parameters are used by the mathematical model to determine the cost of a given XPath query. We compare the performance of our model with existing approaches. Furthermore, we present a use case for an online stream-querying system. The system uses our performance predicate model to estimate the cost for a given XPath query in terms of time/memory. Moreover, it provides an accurate answer for the query's sender. This use case illustrates the practical advantages of performance management with our techniques

Denna forskningsansats undersöker genomförbarheten i att prediktera lavinfara med hjälp av ma-skininlärning i form avXGBoostoch väderdata. Lavinprognoser och meterologisk vädermodelldata harsamlats in för de sex svenska fjällområden där Naturvårdsveket genomlavinprognoser.sepublicerar lavin-prognoser. Lavinprognoserna har hämtats frånlavinprognoser.seoch den vädermodelldata som användsär hämtad från prognosmodellen MESAN, som produceras och tillhandahålls av Sveriges meteorologiskaoch hydrologiska institut. 40 modeller av typenXGBoosthar sedan tränats på denna datamängd, medsyfte att prediktera olika aspekter av en lavinprognos och den övergripande lavinfaran. Resultaten visaratt det möjligt att prediktera den dagligalavinfaranunder säsongen 2018/19 i Södra Jämtlandsfjällenmed en träffsäkerhet på 71% och enmean average errorpå 0,295, genom att applicera maskininlärningpå väderleken för det området. Värdet avXGBoosti sammanhanget har styrkts genom att jämföradessa resultat med resultaten från den enklare metoden logistisk regression, vilken uppvisade en sämreträffsäkerhet på 56% och enmean average errorpå 0,459. Forskningsansatsens bidrag är ett ”proof ofconcept” som visar på genomförbarheten av att med hjälp av maskininlärning och väderdata predikteralavinprognoser. / This research project examines the feasibility of using machine learning to predict avalanche dangerby usingXGBoostand openly available weather data. Avalanche forecasts and meterological modelledweather data have been gathered for the six areas in Sweden where Naturvårdsverket throughlavin-prognoser.seissues avalanche forecasts. The avanlanche forecasts are collected fromlavinprognoser.seand the modelled weather data is collected from theMESANmodel, which is produced and providedby the Swedish Meteorological and Hydrological Institute. 40 machine learning models, in the form ofXGBoost, have been trained on this data set, with the goal of assessing the main aspects of an avalan-che forecast and the overall avalanche danger. The results show it is possible to predict the day to dayavalanche danger for the 2018/19 season inSödra Jämtlandsfjällenwith an accuracy of 71% and a MeanAverage Error of 0.256, by applying machine learning to the weather data for that region. The contribu-tion ofXGBoostin this context, is demonstrated by applying the simpler method ofLogistic Regressionon the data set and comparing the results. Thelogistic regressionperforms worse with an accuracy of56% and a Mean Average Error of 0.459. The contribution of this research is a proof of concept, showingfeasibility in predicting avalanche danger in Sweden, with the help of machine learning and weather data.

Realizou-se neste trabalho uma investigação comparativa do comportamento dos aços ferramenta H13 e THYROTHERM 2999 EFS SUPRA, destinados à fabricação de matrizes para conformação a quente, quando submetidos à fadiga de baixo ciclo a altas temperaturas (FBCAT). A partir de suas curvas de revenimento, foram definidas três durezas de trabalho para cada material (durezas de 42, 52 e 58 HRC), correspondendo a três temperaturas de revenimento distintas e três condições de estudo, buscando-se a condição ótima apresentada por estes materiais para este tipo de aplicação, visando-se então analisar a influência da dureza inicial do material na vida do componente. Foi determinada também a temperatura de ensaio de fadiga isotérmica, em 400°C, correspondente à temperatura de utilização da matriz, ou seja, uma temperatura crítica típica que a matriz atinge durante a solicitação em trabalho. A seguir foram realizados para cada material os ensaios de tração a temperatura ambiente, e na seqüência, os ensaios de tração na temperatura de trabalho definida, que permitiram a determinação dos primeiros parâmetros monotônicos dos materiais, dentre eles uma previsão para os níveis de deformação a serem utilizados nos ensaios de fadiga (0.5,0.6,0.7,0.8,0.9,1.0 e 1.1%), e demais parâmetros como E, k, n, σe, σ’f, ε’f, b, c, que permitiram a elaboração de curvas ε−N, com um modelo estimativo já existente. Finalmente, foram então realizados os ensaios de fadiga isotérmica de baixo ciclo, à temperatura de 400°C, e os resultados foram utilizados para a elaboração das curvas ε−N, resultando então na proposta de um modelo de previsão de resistência à fadiga específico para os materiais pesquisados. / It was made in this work an investigative comparison of the behavior of the tool steels H13 and THYROTHERM 2999 EFS SUPRA, designed for die steels for hot forming, when exposed to high temperature low cycle fatigue (HTLCF). From their tempering curves three material working hardness were defined for each material (hardness of 42, 52 and 58 HRC), corresponding to three different tempering temperatures, and so three study cases for each material, searching for the best condition for this kind of application, and to assess the influence of the initial hardness on the part material life. The isothermal low cycle fatigue test temperature was either defined at 400°C, corresponding to the used temperature at the die steel, i.e., a critical typical temperature that the forging dies reach on hot working. After that, tensile tests were performed for both materials, at room temperature, and at the working temperature formerly defined, and these tests allowed the definition of the first monotonic parameters for these materials, among them predictions for strain levels (0.5, 0.6, 0.7, 0.8, 0.9, 1.0 and 1.1%), to be used on fatigue tests, and further parameters like E, k, n, σe, σ’f, ε’f, b, c, that allowed the elaboration of ε−N curves, based on a still existing prediction model. Finally, isothermal low cycle fatigue tests were performed, at 400°C, and the results were used for ε−N curves elaboration, resulting on a prediction model of the fatigue strength specified for the assessed materials.

The present study focused on evaluating some crash prediction models for two lane highways on Brazilian conditions. Also, the transferability of models was considered, specifically by means of a comparison between Brazil, HSM and Florida. The analysis of two lane highways crash prediction models was promising when the road characteristics were well known and there was not much difference from base conditions. This conclusion was attained regarding the comparison of results for all segments, non-curved segments and curved segments, confirming that a transferred model can be used with caution. In addition, two novel models for Brazilian two-lane highways segments were estimated. The model developed showed better results for non-curved segments in the calibration/validation sample. Thus, for a general analysis purpose of non-curved segments this model is recommended. Finally, there are many factors that could not be measured by these models and reflects road safety various condition. Even so, the study of crash predict models in Brazilian context could provide a better start point in safety road analysis. / O foco desta pesquisa foi avaliar a aplicação de alguns modelos de previsão de acidentes em rodovias de pista simples de três estados brasileiros. Ainda, a transferabilidade destes modelos foi abordada, especificamente por meio de uma comparação entre características do Brasil, Florida e aquelas recomendadas pelo Highway Safety Manual. O uso dos distintos modelos se mostrou promissor para situações nas quais as características da via se mantiveram semelhantes às condições para as quais os modelos foram desenvolvidos. A avaliação foi empreendida para todos os segmentos homogêneos, separados posteriormente segundo a existência de curvas horizontais. Adicionalmente, dois novos modelos foram equacionados para a amostra brasileira. O modelo de previsão de acidentes desenvolvido apresentou melhores medidas de desempenho para segmentos sem curvas horizontais, sendo recomendável para previsão de acidentes em análises preliminares. Por fim, foi constatado que outros fatores não contemplados pelos modelos podem ter impactado as condições de segurança dos locais estudados. Ainda assim, essa pesquisa representa no contexto do Brasil um ponto de partida em análises relacionadas à segurança de rodovias de pista simples.

La maladie de Parkinson est l'une des principales pathologies neurodégénératives, mais il existe peu de données françaises sur sa fréquence. Des études antérieures suggèrent que l'exposition aux pesticides est un facteur de risque de maladie de Parkinson ; la population agricole représente donc une population particulièrement intéressante. Nous avons étudié l’intérêt d’une méthode utilisant les remboursements de médicaments antiparkinsoniens pour identifier les patients parkinsoniens parmi les affiliés de la Mutualité Sociale Agricole dans cinq départements en 2007 en développant, à partir de 1 114 consommateurs de médicaments antiparkinsoniens, un modèle prédictif dont nous avons estimé les performances (sensibilité =92,5 %, spécificité = 86,4 %, statistique c = 0,953 %). A partir de ce modèle, nous estimons que la prévalence de la maladie de Parkinson dans les cinq départements est de 6,20 ‰ après18 ans et de 2,76 ‰ après standardisation sur l’âge et le sexe (population française de 2007comme référence). La prévalence est 1,3 fois plus élevée dans les cantons où la densité en exploitations spécialisées en vergers et autres cultures permanentes est la plus élevée ; ce typed’exploitation est caractérisé par une utilisation intensive de pesticides, notamment d’insecticides. Des informations détaillées sur l’utilisation professionnelle de pesticides ont été recueillies dans une étude cas-témoins (331 cas, 660 témoins). Nous observons une association entre la maladie de Parkinson et l’utilisation intensive (nombre élevéd’applications par an) de fongicides et d’insecticides. Parmi les différents types d’exploitations, une association avec la maladie de Parkinson est uniquement retrouvée pour l’utilisation de pesticides dans les exploitations viticoles. Enfin, l’association avec l’exposition professionnelle aux pesticides semble plus marquée pour la présentation clinique de la maladie caractérisée par la présence d'un tremblement de repos / Parkinson's disease is one the main neurodegenerative diseases, but there are little French data on its frequency. Previous studies suggest that pesticide exposure is a risk factor for Parkinson's disease; the agricultural population is therefore particularly interesting. Westudied the value of a method using antiparkinsonian drugs claims to identify Parkinson's disease patients among members of the Mutuality Sociale Agricole in five districts in 2007 by developing, based on 1,114 antiparkinsonian drugs users, a predictive model and estimatingits performances (sensitivity = 92.5%, specificity = 86.4%, c-statistic = 0.953). Based on thismodel, we estimate that the prevalence of Parkinson's disease in the five districts is of 6.20 ‰after 18 years and 2.76 ‰ after standardization on the age- and sex-distribution of the 2007 French population. The prevalence is 1.3 times higher in cantons where the density of farms specialized in orchards and other permanent crops is highest; this activity is characterized by intensive use of pesticides, especially insecticides. Detailed information on professional useof pesticides were collected in a case-control study (331 cases, 660 controls). We observed anassociation between Parkinson's disease and intensive use (high number of applications peryear) of fungicides and insecticides. Among different types of farms, an association with Parkinson's disease is only found for pesticide use in vineyards. Finally, the association with occupational exposure to pesticides seems more pronounced for the clinical presentation of the disease characterized by the presence of resting tremor.

「企業財務危機」向來為銀行及企業體所關注的重大課題，學術界對於此危機預警模型之研究篇幅亦不少，惟實務上卻甚少有金融機構將此等模型實際應用於授信決策。歸納其原因主要為樣本收集與統計方法運用的配合問題。本研究主要目的即擬藉廣泛蒐集銀行「逾期」與「正常」往來企業授信案，分別業別，分別建立一套能實際運用於銀行授信決策之企業財務危機預警模型。 　　本文以問卷收集省屬七行庫授信案件為樣本，其中包括紡織工業、金屬製品業、電工器材業、進出口貿易業及建築材料業等五個行業171家逾期公司及配對之171家正常公司，以其各項財務比率先作常態分配檢定，再分行業別分別比較逾期公司與正常公司財務特質有無差異，然後運用逐步迴歸分析法萃取關鍵變數，建立各行業之Logit迴歸預警模型。最後，確立各模型之臨界點，測試各模型之區別能力及預測能力，並比較整體公司模型與各行業別模型之預測能力。 　　研究結論： 　　（一）銀行授信戶財務比率分配均違反常態性。 　　（二）逾期公司相對於正常公司大部份財務特質均有顯著差異，且逾期公司具有較低變現性，高的財務槓桿性，低的經營能力及低的獲利力。 　　（三）本研究各行業別預警模型之預測正確率分別為：紡織工業88.24%、金屬製品業90%、電工器材業90.48%、進出口貿易業88.89%、建築材料業90%。 　　（四）最後以各行業之後期樣本測試整體公司模型，發現除金屬製品業及進出口貿易業其行業別預警模型與整體公司預警模型之預測正確率相同外，餘三種行業預警模型之預測正確率均較整體公司模型為高。顯示為有效提昇銀行授信財務危機預警模型的預測能力，依行業別分別建立模型，實有必要。

國內企業失敗的研究大都定位在財務危機的領域，且幾乎完全以股票上市 公司為研究對象。因受限於國內證券市場規模太小，研究樣本受到相當大 的限制，歷年來發生財務危機的案例甚少，所以財務預警系統只說明模式 本身區別成效，至於樣本外的測試，則付之闕如，無法了解此一模式對不 同時期的樣本是否仍有相同的區別能力，模式穩定性也就無從得知了。本 研究採用洪榮華（民82）的研究，擴大解釋危機的定義，以企業發生虧損 為一觀察事件，找尋足夠的樣本來建構財務危機預測模式，並從事樣本外 的預測，以驗證區別模式的穩定性。前人的研究結果顯示：大多數的預測 模式在樣本內的測試均有相當不錯的區別能力，樣本外的預測能力卻大幅 下降。Platt & Platt（1990）對於模式不穩定的原因歸納出兩個結論： 財務報表資料會隨時間的經過產生不穩定的現象、產業間財務資料的差異 ，經由產業相對財務比率的調整可以消除產業效果使得模式的區別能力趨 於穩定。Marquette（1980） 指出，良好的預測模型必須具有動態模型、 經得起時間考驗、以及不受時間約束等特性。也就是說，一個優良的財務 預警模型不會受到產業與時間因素的影響，可以適用在不同產業、時期， 即模式具有高度的穩定性。因此，基於上述理由本研究之目的有三：1.比 較以產業相對財務比率與公司財務比率所建立的預警模式對企業財務危機 的預測，何者有較佳的區別能力。2.驗證由原始樣本所建立區別模式對同 一時期的保留樣本，是否有顯著的區別能力，用以了解模式是否有同期的 適用性。3.驗證由原始樣本所建立區別模式對後期樣本，是否有顯著的預 測能力，用以了解模式是否有跨期的適用性。結論有四：1.公司財務比率 模型並非一穩定模型。模式本身具有區別能力，但卻不具有同期的適用性 與跨期的適用性。2.相對而言，產業相對財務比率模型為一穩定模型。模 式本身不僅具有區別能力，也同時具有跨期的適用性。3.就模式的區別能 力與預測能力而言，除危機前一年外，產業相對財務比率模型的區別效果 ，並未顯著優於公司財務比率模型。4.從本研究中發現，無論公司或產業 相對比率模型，對財務危機事件的預測，短期的預測能力高，長期則顯著 下降。

會計保守性為盈餘品質特性之一，其已存在好幾個世紀，且有證據顯示在過去三十年間有增加的趨勢，對於會計實務影響深遠。會計保守性會造成資產的帳面價值被低估，但卻提高會計盈餘品質。 本研究探討會計保守程度與財務困難機率之關係、機構投資人對於公司之持股與會計保守性之關係以及財務困難機率與機構投資人持股之關係。實證結果發現：會計保守程度較高之公司其財務危機機率較低，顯示會計保守程度較高之公司其財務體質較佳；公司財務困難機率愈低，則機構投資人持股愈多；而會計保守程度對於機構投資人之持股行為亦有顯著之影響。 / Conservatism is an attribute of accounting earnings, which has been present in accounting practice for centuries and appears to be more popular in the last 30 years. It persistently understates the book value of net assets while increases earnings quality. The influence of conservatism on accounting is significant. This study examines the relations among accounting conservatism, financial distress, and institutional holding. The findings indicate that the higher the accounting conservatism the lower the financial distress probability, the lower the financial distress probability the more the institutional holding, and the accounting conservatism has also significant effect on institutional holding.

In this thesis, we design, develop and implement parallel algorithms for irregular problems on heterogeneous multi-core architectures. Irregular problems exhibit random and unpredictable memory access patterns, poor spatial locality and input dependent control flow. Heterogeneous multi-core processors vary in: clock frequency, power dissipation, programming model (MIMD vs. SIMD), memory design and computing units, scalar versus vector units. The heterogeneity of the processors makes designing efficient parallel algorithms for irregular problems on heterogeneous multicore processors challenging. Techniques of mapping tasks or data on traditional parallel computers can not be used as is on heterogeneous multi-core processors due to the varying hardware. In an attempt to understand the efficiency of futuristic heterogeneous multi-core architectures on applications we study several computation and bandwidth oriented irregular problems on one heterogeneous multi-core architecture, the IBM Cell Broadband Engine (Cell BE). The Cell BE consists of a general processor and eight specialized processors and addresses vector/data-level parallelism and instruction-level parallelism simultaneously. Through these studies on the Cell BE, we provide some discussions and insight on the performance of the applications on heterogeneous multi-core architectures. Verifying these experimental results require some performance modeling. Due to the diversity of heterogeneous multi-core architectures, theoretical performance models used for homogeneous multi-core architectures do not provide accurate results. Therefore, in this thesis we propose an analytical performance prediction model that considers the multitude architectural features of heterogeneous multi-cores (such as DMA transfers, number of instructions and operations, the processor frequency and DMA bandwidth). We show that the execution time from our prediction model is comparable to the execution time of the experimental results for a complex medical imaging application.

This thesis takes place in the field of Ambient Intelligence (AmI). AmI Systems are interactive systems composed of many heterogeneous components. From a hardware perspective these components can be divided into two main classes: sensors, using which the system observes its surroundings, and actuators, through which the system acts upon its surroundings in order to execute specific tasks.From a functional point of view, the goal of AmI Systems is to activate some actuators, based on data provided by some sensors. However, sensors and actuators may suffer failures. Our motivation in this thesis is to equip ambient systems with self fault detection capabilities. One of the particularities of AmI systems is that instances of physical resources (mainly sensors and actuators) are not necessarily known at design time; instead they are dynamically discovered at run-time. In consequence, one could not apply classical control theory to pre-determine closed control loops using the available sensors. We propose an approach in which the fault detection and diagnosis in AmI systems is dynamically done at run-time, while decoupling actuators and sensors at design time. We introduce a Fault Detection and Diagnosis framework modeling the generic characteristics of actuators and sensors, and the physical effects that are expected on the physical environment when a given action is performed by the system's actuators. These effects are then used at run-time to link actuators (that produce them) with the corresponding sensors (that detect them). Most importantly the mathematical model describing each effect allows the calculation of the expected readings of sensors. Comparing the predicted values with the actual values provided by sensors allows us to achieve fault-detection.