Machine Translation (MT) for language pairs with long distance dependencies and word reordering, such as German-English, is prone to producing output that is lexically or syntactically incoherent. Statistical MT (SMT) models used explicit or latent syntax to improve reordering, however failed at capturing other long distance dependencies. This thesis explores how explicit sentence-level syntactic information can improve translation for such complex linguistic phenomena. In particular, we work at the level of the syntactic-semantic interface with representations conveying the predicate-argument structures. These are essential to preserving semantics in translation and SMT systems have long struggled to model them. String-to-tree SMT systems use explicit target syntax to handle long-distance reordering, but make strong independence assumptions which lead to inconsistent lexical choices. To address this, we propose a Selectional Preferences feature which models the semantic affinities between target predicates and their argument fillers using the target dependency relations available in the decoder. We found that our feature is not effective in a string-to-tree system for German-English and that often the conditioning context is wrong because of mistranslated verbs. To improve verb translation, we proposed a Neural Verb Lexicon Model (NVLM) incorporating sentence-level syntactic context from the source which carries relevant semantic information for verb disambiguation. When used as an extra feature for re-ranking the output of a German-English string-to-tree system, the NVLM improved verb translation precision by up to 2.7% and recall by up to 7.4%. While the NVLM improved some aspects of translation, other syntactic and lexical inconsistencies are not being addressed by a linear combination of independent models. In contrast to SMT, neural machine translation (NMT) avoids strong independence assumptions thus generating more fluent translations and capturing some long-distance dependencies. Still, incorporating additional linguistic information can improve translation quality. We proposed a method for tightly coupling target words and syntax in the NMT decoder. To represent syntax explicitly, we used CCG supertags, which encode subcategorization information, capturing long distance dependencies and attachments. Our method improved translation quality on several difficult linguistic constructs, including prepositional phrases which are the most frequent type of predicate arguments. These improvements over a strong baseline NMT system were consistent across two language pairs: 0.9 BLEU for German-English and 1.2 BLEU for Romanian-English.

Orientador: Eliseu Savério Sposito / Banca: Everaldo Santos Melazzo / Banca: Arthur Magon Whitacker / Banca: Saint-Clair Cordeiro da Trindade Júnior / Banca: Sandra Lencioni / Resumo: As implicações espaciais da instalação de grandes projetos na Amazônia, como estradas, usinas hidrelétricas e indústrias de mineração, têm sido objeto de várias frentes de análise em Geografia no intuito de debater a respeito do papel transformador desses empreendimentos em contextos urbanos particulares. A partir de 2010, quando é liberada a licença de instalação da usina hidrelétrica de Belo Monte, percebe-se uma dinâmica social diferenciada, sobretudo marcada pela produção de novas formas espaciais e novas relações que envolvem a cidade e a rede urbana. Considerando essa realidade, a presente pesquisa resolveu articular o processo de construção da usina de Belo Monte ao tema das cidades médias, tendo como como base a seguinte questão: em que medida as transformações espaciais desencadeadas a partir da instalação da usina hidrelétrica de Belo Monte produzem uma reestruturação da cidade e uma reestruturação urbana? A área efetiva de estudos abrange, portanto, duas escalas de análise articuladas: a cidade de Altamira-PA, situada às margens do rio Xingu, a 54 km² de distância da área de construção da usina de Belo Monte, e a Região de Influência de Altamira (RIA). Esta última é baseada em dois estudos complementares do IBGE (REGIC de 2007 e Divisão Urbano-regional, de 2013) e corresponde aos centros urbanos situados no interfluxo entre o rio Xingu (Vitória do Xingu, Senador José Porfírio e Porto de Moz) e a rodovia Transamazônica (Altamira, Anapu, Pacajá, Medicilândia Uruará e... (Resumo completo, clicar acesso eletrônico abaixo) / The spatial implications of large projects installation in the Amazon, such as roads, hydroelectric plants and mining industries have been the subject of several fronts of analysis in geography in order to demonstrate the transformative role of these enterprises in particular urban contexts. Starting in 2010, when the license installation of the hydroelectric plant of Belo Monte is released, it is clear a different social dynamic, particularly marked by the production of new spatial forms and new relationships involving the city and the urban network. Given this reality, this study decided to articulate the process of construction of the Belo Monte plant to the issue of medium-sized cities, based on the following question: how far the spatial transformations triggered from the installation of the hydroelectric plant of Belo Monte produce a city reorganization and an urban reorganization? The effective area of study covers, therefore, two articulated scales of analysis: the city of Altamira (Pará), situated at the margins of the Xingu river - 54km² away from the area of construction of the Belo Monte plant -, and the Region of Influence in Altamira (RIA). This latter is based on two complementary studies of IBGE (REGIC of 2007 and Division Urban-Regional, of 2013) and corresponds to the urban centers in the interflow between the Xingu river (Vitória do Xingu, Senador José Porfírio and Porto de Moz) and the Trans-Amazonian highway (Altamira, Anapu, Pacajá, Medicilândia, Uruará and Brasil Novo). With regard to Altamira, it is started from the idea that this city serves as a medium-sized city, since it functions as an intermediary between the big cities and local centers, allowing to enter authors who work in this field of study... (Complete abstract click electronic access below) / Doutor

Este estudo foi realizado com o objetivo de conhecer e analisar como o Núcleo de Tecnologia Educacional de Terra Nova do Norte está capacitando os professores da rede pública de ensino em Informática da Educação; como essa capacitação é construída nos Laboratórios de Informática nas escolas; e como os professores estão fazendo uso do computador como ferramenta pedagógica. Pretendemos determinar quais são as dificuldades encontradas nesse tipo de capacitação, estabelecendo quais as reais perspectivas dos educadores e dos alunos do ensino fundamental; à luz dos resultados, elaboraremos um conjunto de sugestões com o intuito de melhorar o nível de qualidade de atuação dos envolvidos com tecnologia da informação e da comunicação. Além disso visitamos os núcleos de: Cuiabá, Cáceres, Diamantino, Nova Xavantina, Sinop e Terra Nova do Norte; os Laboratórios de Informática nos municípios de: Terra Nova do Norte, Colider, Itaúba, Claudia, Marcelândia, Feliz Natal e Sinop. Também visitamos os professores de sala de aula que atuam no Ensino Fundamental nessas escolas. Com dificuldades financeiras, sem meios de transporte próprio, com falta de pessoal técnico e especializado e com acumulo de funções e, conseqüentemente, sobrecarga de atividades, o Núcleo de Tecnologia Educacional de Terra Nova do Norte, através de seus especialistas, vem realizando um trabalho exaustivo, mas sem perspectiva: os cursos de capacitação relâmpagos não preparam os professores nem para funções elementares, muitos menos para funções complexas Para esses cursos, deve-se conceber uma carga horária mínima de 100 (Cem) horas/aula, resguardando a relação indissociável entre teoria e prática. A capacitação dos professores multiplicadores, que ainda acontece em nível básico, é a grande dificuldade que os especialistas do núcleo estão encontrando nesse processo, pois não há como dar uma boa capacitação em informática educacional se a maioria dos professores nunca sentaram em frente a um computador. Há necessidade de se rever com muito carinho esse processo de informatização das escolas públicas, tanto na dimensão epistemológica quanto na profissionalização. Há um grande compromisso e força de vontade do pessoal das escolas e dos núcleos, o que nos leva a afirmar que, se não fosse pela tenacidade desses professores o programa já estaria em fase terminal. O programa, portanto, necessita de mais apoio e de investimento em suas potencialidades técnica e pedagógica, pois do modo como está, pouco contribui para a democratização e a melhoria do ensino.

Tese de mestrado integrado. Engenharia Electrotécnica e de Computadores. Área de Especialização de Energia. Faculdade de Engenharia. Universidade do Porto. 2012

Les aminoacyl-ARNt synthétases mitochondriales humaines (aaRS mt) sont des enzymes clés de la traduction mitochondriale. Elles catalysent l'aminoacylation des ARNt par les acides aminés correspondent. Des mutations dans leurs gènes sont corrélées à des pathologies avec un large spectre de phénotypes cliniques, mais aux mécanismes moléculaires sous-jacents encore incompris. L'objectif de ce travail de thèse s'intègre dans les axes scientifiques du laboratoire, mais élargit l'intérêt et les connaissance à un système encore peu exploré: l'arginyl-ARNt synthétase mitochondriale (ArgRS mt). Des mutations dans la ArgRS sont liées à une hypoplasie Pontocérébelleuse (PCH6), une pathologie neurodéveloppementale sévère. Le travail de cette thèse s’articule autour de 3 axes : (I) L’analyse des phénotypes cliniques des pathologies liées aux mutations dans les aaRS mt, (II) La caractérisation des propriétéscellulaires de l’ArgRS mt, et (III) L'étude de l’impact de mutations « pathologiques » sur diverses propriétés de l’ArgRS mt. Combinés avec les travaux précédents, les résultats obtenus sont une contribution importante à l'élargissement des connaissances fondamentales des mt aaRSs, et apportent un nouvel éclairage sur le lien entre les mt-aaRSs-mutations et la maladie. / Human mitochondrial aminoacyl-tRNA synthetases (mt-aaRSs) are housekeeping enzymes involved in the mitochondrial translation. They catalyze the aminoacylation of tRNAs with their cognate amino acids. Mutations in their nuclear genes are correlated with pathologies with a broad spectrum of clinical phenotypes, but with so far no clear explanations about the underlying molecular mechanism(s). The aim of this PhD work follows the long-standing efforts of the host laboratory but expand the interest and knowledge to an unexplored system: the human mitochondrial arginyl-tRNA synthetase (mt-ArgRS). Mutations in the mt-ArgRS lead to Pontocebellar hypoplasia type 6, a severe neuro-developmental pathology. I thus contributed to i) comprehensively analyze the clinical data reported in pathologies related to mutations on mt-aaRSs, resulting in a categorization according to the affected anatomical system; ii) decipher some cellular properties of the mt-ArgRS; and iii) investigate to impact of disease-associated mutations on mt-aaRSs properties. Combined with previous works, the present results expand the knowledge of the mt-aaRSs, shedding new light on the link between mt-aaRSs-mutations and disease.

Realistic probabilistic hazard forecasts for re-awakening volcanoes rely on making an accurate estimation of their past eruption frequency and magnitude for a period long enough to view systematic changes or evolution. Adding an in-depth knowledge of the local underlying magmatic or tectonic driving processes allows development of even more robust eruption forecasting models. Holocene tephra records preserved within lacustrine sediments and soils on and surrounding the andesitic stratovolcano of Mt. Taranaki (Egmont Volcano), New Zealand, were used to 1) compile an eruption catalogue that minimises bias to carry out frequency analysis, and 2) identify magmatic processes responsible for variations in activity of this intermittently awakening volcano. A new, highly detailed eruption history for Mt. Taranaki was compiled from sediment sequences containing Holocene tephra layers preserved beneath Lakes Umutekai and Rotokare, NE and SE of the volcano’s summit, respectively, with age control provided by radiocarbon dating. To combine the two partly concurrent tephra records both geochemistry (on titanomagnetite) and statistical measures of event concurrence were applied. Similarly, correlation was made to proximal pyroclastic sequences in all sectors around the 2518 m-high edifice. This record was used to examine geochemical variations (through titanomagnetite and bulk chemistry) at Mt. Taranaki in unprecedented sampling detail. To develop an unbiased sampling of eruption event frequency, a technique was developed to distinguish explosive, pumice-forming eruptions from dome-forming events recorded in medial ash as fine-grade ash layers. Recognising that exsolution lamellae in titanomagnetite result from oxidation processes within lava domes or plugs, their presence within ash deposits was used to distinguish falls elutriated from blockand- ash flows. These deposits are focused in particular catchments and are hence difficult to sample comprehensively. Excluding these events from temporal eruption records, the remaining, widespread pumice layers of sub-plinian eruptions at a single site of Lake Umutekai presented the lowest-bias sampling of the overall event frequency. The annual eruption frequency of Mt. Taranaki was found to be strongly cyclic with a 1500-2000 year periodicity. Titanomagnetite, glass and whole-rock chemistry of eruptives from Mt. Taranaki’s Holocene history all display distinctive compositional cycles that correspond precisely with the event frequency curve for this volcano. Furthermore, the largest known eruptions from the volcano involve the most strongly evolved magmas of their cycle and occur during the eruptive-frequency minimum, preceding the longest repose intervals known. Petrological evidence reveals a two-stage system of magma differentiation and assembly operating at Mt. Taranaki. Each of the identified 1500-2000 year cycles represent isolated magma batches that evolved at depth at the base of the crust before periodically feeding a mid-upper crustal magma storage system.

Ce travail de thèse a été consacré à la mise au point d'une nouvelle microscopie électrochimique à force atomique (AFM-SECM) à haute résolution, totalement innovante, désignée par l'acronyme MT/AFM-SECM pour « Mediator Tethered–AFM/SECM », et pour laquelle le médiateur rédox ferrocène (Fc) est lié à la micro-électrode sonde combinée par le biais de chaînes flexibles polyéthylène glycol (PEG). Il a été établi que les têtes ferrocènes ainsi attachées à l'extrémité de la pointe-sonde AFM-SECM sont capables de sonder la réactivité locale d'un substrat. Il a de plus été démontré que les pointes AFM-SECM Fc-PEGylées peuvent être utilisées en imagerie en mode tapping, permettant l'acquisition simultanée d'un courant de feedback électrochimique et de la topographie du substrat, et ce avec une résolution latérale et verticale de l'ordre du nanomètre. Ce nouveau type de microscopie SECM, devrait trouver de nombreuses applications en partie parce qu'il permet d'obtenir une résolution SECM de l'ordre du nanomètre, dictée par la longueur des chaînes PEG, sans avoir à utiliser des sondes nanométriques. Cette nouvelle microscopie présente donc une résolution suffisante pour permettre de localiser des macromolécules individuelles, en particulier des molécules d'enzyme rédox, sur une surface. Mais de plus, parce qu'elle est libre des contraintes diffusionnelles de la SECM « classique », la microscopie MT/AFM-SECM devrait, en principe, permettre également de caractériser le fonctionnement cinétique de ces enzymes, molécule d'enzyme par molécule d'enzyme.

<p>The resolution power of single Very Low Frequency (VLF) data and multi-frequency Radiomagnetotelluric (RMT) data in delineating conductive structures typical for the sedimentary cover and crystalline basement in Scandinavia is studied with a view to future developments of the technique to increasing the frequency range into the LW radio band. Airborne and ground VLF data are interpreted and correlated with RMT measurements made on the ground to better understand the resolution power of VLF data. To aid in this understanding single and multifrequency VLF and RMT responses for some typical resistivity structures are analyzed. An analytic model is presented for obtaining unique transfer functions from measurements of the electromagnetic components on board an air-plane or on the ground. Examples of 2D inversion of ground and airborne VLF profiles in Sweden are shown to demonstrate the quantitative interpretation of VLF data in terms of both lateral and depth changes of the resistivity in the uppermost crust.</p><p>Geothermal resources are ideal targets for Electromagnetic (EM) methods since they produce strong variations in underground electrical resistivity. Modelling of Magnetotelluric (MT) data in SW Iceland indicates an alteration zone beneath the surface, where there are no obvious geothermal manifestations, in between Hengill and Brennisteinsfjoll geothermal systems. It suggests that a hydrothermal fluid circulation exists at depth. It also proves that the MT method, with its ability to map deep conductive features can play a valuable role in the reconnaissance of deep geothermal systems in active rift regimes such as in Iceland.</p><p>A damped nonlinear least-squares inversion approach is employed to invert Controlled Source Tensor MT (CSTMT) data for azimuthal anisotropy in a 1D layered earth. Impedance and tipper data are inverted jointly. The effects of near-surface inhomogeneities are parameterized in addition to each layer parameter(s). Application of the inversion algorithm to both synthetic and field data shows that the CSTMT method can be used to detect azimuthal anisotropy under realistic conditions with near surface lateral heterogeneities.</p>

Multithreaded processors are now common in the industry as they offer high performance at a low cost. Traditionally, in such processors, the assignation of hardware resources between the multiple threads is done implicitly, by the hardware policies. However, a new class of multithreaded hardware allows the explicit allocation of resources to be controlled or biased by the software. Currently, there is little or no coordination between the allocation of resources done by the hardware and the prioritization of tasks done by the software.This thesis targets to narrow the gap between the software and the hardware, with respect to the hardware resource allocation, by proposing a new explicit resource allocation hardware mechanism and novel schedulers that use the currently available hardware resource allocation mechanisms.It approaches the problem in two different types of computing systems: on the high performance computing domain, we characterize the first processor to present a mechanism that allows the software to bias the allocation hardware resources, the IBM POWER5. In addition, we propose the use of hardware resource allocation as a way to balance high performance computing applications. Finally, we propose two new scheduling mechanisms that are able to transparently and successfully balance applications in real systems using the hardware resource allocation. On the soft real-time domain, we propose a hardware extension to the existing explicit resource allocation hardware and, in addition, two software schedulers that use the explicit allocation hardware to improve the schedulability of tasks in a soft real-time system.In this thesis, we demonstrate that system performance improves by making the software aware of the mechanisms to control the amount of resources given to each running thread. In particular, for the high performance computing domain, we show that it is possible to decrease the execution time of MPI applications biasing the hardware resource assignation between threads. In addition, we show that it is possible to decrease the number of missed deadlines when scheduling tasks in a soft real-time SMT system.

農林水産研究情報センターで作成したPDFファイルを使用している。