Thèses sur le sujet « Marginal parameters »

The graphical models (GM) for categorical data are models useful to representing conditional independencies through graphs. The parametric marginal models for categorical data have useful properties for the asymptotic theory. This work is focused on nding which GMs can be represented by marginal parametrizations. Following theorem 1 of Bergsma, Rudas and Németh [9], we have proposed a method to identify when a GM is parametrizable according to a marginal model. We have applied this method to the four types of GMs for chain graphs, summarized by Drton [22]. In particular, with regard to the so-called GM of type II and GM of type III, we have found the subclasses of these models which are parametrizable with marginal models, and therefore they are smooth. About the so-called GM of type I and GM of type IV, in the literature it is known that these models are smooth and we have provided new proof of this result. Finally we have applied the mean results concerning the GM of type II on the EVS data-set.

Submitted by Bernardo Andrade Lyrio Modenesi (bmodenesi@gmail.com) on 2015-07-22T21:31:40Z No. of bitstreams: 1 modenesi_thesis_v4whatseparatemen.pdf: 819403 bytes, checksum: d0a2f018fe44526173ae5b3f9e4c13a9 (MD5)
Rejected by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br), reason: Prezado Bernardo bom dia, Nome da fundação e escola em negrito, o trabalho é em inglês vem primeiro o ABSTRACT depois o RESUMO e para aprovar o trabalho é necessário o seu orientador encaminhar e-mail para o cmcd@fgv.br com o novo Título autorizando o protocolo da versão final. Aguardo! Grata. Suzi 3799-7876 on 2015-07-23T11:30:51Z (GMT)
Submitted by Bernardo Andrade Lyrio Modenesi (bmodenesi@gmail.com) on 2015-07-23T14:56:25Z No. of bitstreams: 1 modenesi_thesis_v4.pdf: 819610 bytes, checksum: 8482a7c46b6c9bada5d33905e41147d7 (MD5)
Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2015-07-23T17:16:31Z (GMT) No. of bitstreams: 1 modenesi_thesis_v4.pdf: 819610 bytes, checksum: 8482a7c46b6c9bada5d33905e41147d7 (MD5)
Made available in DSpace on 2015-07-23T18:29:24Z (GMT). No. of bitstreams: 1 modenesi_thesis_v4.pdf: 819610 bytes, checksum: 8482a7c46b6c9bada5d33905e41147d7 (MD5) Previous issue date: 2015-06-25
When estimating policy parameters, also known as treatment effects, the assignment to treatment mechanism almost always causes endogeneity and thus bias many of these policy parameters estimates. Additionally, heterogeneity in program impacts is more likely to be the norm than the exception for most social programs. In situations where these issues are present, the Marginal Treatment Effect (MTE) parameter estimation makes use of an instrument to avoid assignment bias and simultaneously to account for heterogeneous effects throughout individuals. Although this parameter is point identified in the literature, the assumptions required for identification may be strong. Given that, we use weaker assumptions in order to partially identify the MTE, i.e. to stablish a methodology for MTE bounds estimation, implementing it computationally and showing results from Monte Carlo simulations. The partial identification we perfom requires the MTE to be a monotone function over the propensity score, which is a reasonable assumption on several economics' examples, and the simulation results shows it is possible to get informative even in restricted cases where point identification is lost. Additionally, in situations where estimated bounds are not informative and the traditional point identification is lost, we suggest a more generic method to point estimate MTE using the Moore-Penrose Pseudo-Invese Matrix, achieving better results than traditional methods.
A estimação de parâmetros relevantes no contexto de intervenções políticas, também conhecidos como efeitos de tratamento, enfrenta diversos problemas práticos como o viés relacionado ao mecanismo de atribuição do status de tratamento. Adicionalmente, efeitos heterogênos na literatura de Avaliação de Impactos é mais comum que efeitos homogêneos. Em situações nas quais estes problemas estão presentes, a estimação do Efeito Marginal de Tratamento (MTE) emprega o método de variáveis instrumentais para contornar o viés de seleção ao tratamento, obtendo ao mesmo tempo efeitos heterogêneos. Mesmo que a literatura identifique pontualmente este parâmetro, as hipóteses por trás da identificação são costumeiramente restritivas. Neste trabalho pretende-se afrouxar as hipóteses vigentes na literatura de modo a obter identificação parcial do MTE, requerendo apenas monotonicidade do mesmo ao longo das diferentes propensões ao tratamento, o que é comum em diversos exemplos da literatura econômica. Simulações de Monte Carlo são performadas, resultando em limites para o MTE que se mostram informativos, mesmo em situações restritas nas quais a tradicional identificação pontual é perdida. Complementarmente, em situações quando a identificação parcial não é informativa e a tradicional ponto identificação é perdida, propomos a ponto identificação utilizando a matrix pseudo inversa de Moore-Penrose. Esta metodologia prescinde da hipótese de monotonicidade e resulta em melhores estimativas quando comparada aos métodos tradicionais.

The Multidimensional Generalized Graded Unfolding Model (MGGUM) is a proximity-based, noncompensatory item response theory (IRT) model with applications in the context of attitude, personality, and preference measurement. Model development used fully Bayesian Markov Chain Monte Carlo (MCMC) parameter estimation (Roberts, Jun, Thompson, & Shim, 2009a; Roberts & Shim, 2010). Challenges can arise while estimating MGGUM parameters using MCMC where the meaning of dimensions may switch during the estimation process and difficulties in obtaining informative starting values may lead to increased identification of local maxima. Furthermore, researchers must contend with lengthy computer processing time. It has been shown alternative estimation methods perform just as well as, if not better than, MCMC in the unidimensional Generalized Graded Unfolding Model (GGUM; Roberts & Thompson, 2011) with marginal maximum a posteriori (MMAP) item parameter estimation paired with expected a posteriori (EAP) person parameter estimation being a viable alternative. This work implements MMAP/EAP parameter estimation in the multidimensional model. Additionally, item location initial values are derived from detrended correspondence analysis (DCA) based on previous implementation of correspondence analysis in the GGUM (Polak, 2011). A parameter recovery demonstrates the accuracy of two-dimensional MGGUM MMAP/EAP parameter estimates and a comparative analysis of MMAP/EAP and MCMC demonstrates equal accuracy, yet much improved efficiency of the former method. Analysis of real attitude measurement data also provides an illustrative application of the model.

Item response theory (IRT) uses a family of statistical models for estimating stable characteristics of items and examinees and defining how these characteristics interact in describing item and test performance. With a focus on the three-parameter logistic IRT (Birnbaum, 1968; Lord, 1980) model, the current study examines the accuracy and variability of the item parameter estimates from the marginal maximum a posteriori estimation via an expectation-maximization algorithm (MMAP/EM) and the Markov chain Monte Carlo Gibbs sampling (MCMC/GS) approach. In the study, the various factors which have an impact on the accuracy and variability of the item parameter estimates are discussed, and then further evaluated through a large scale simulation. The factors of interest include the composition and length of tests, the distribution of underlying latent traits, the size of samples, and the prior distributions of discrimination, difficulty, and pseudo-guessing parameters. The results of the two estimation methods are compared to determine the lower limit--in terms of test length, sample size, test characteristics, and prior distributions of item parameters--at which the methods can satisfactorily recover item parameters and efficiently function in reality. For practitioners, the results help to define limits on the appropriate use of the BILOG-MG (which implements MMAP/EM) and also, to assist in deciding the utility of OpenBUGS (which carries out MCMC/GS) for item parameter estimation in practice.

The aim of this master thesis is to verify possibility to use Poisson distribution for predicting soccer matches. At first for analysis is applied the original model from English statisticians Mark J. Dixon and Stuart G. Coles from 1997. Thereafter the model is extended in the thesis. All models are based on the maximum likelihood method. Chosen league for deducing conclusions is the first English league - Premier League. The matches are played in the period from season 2004/2005 to half of season 2015/2016. For identification of models performance are used the most market odds from American bookmaker Pinnacle. In the theoretical part are described models and statistical methods that are used in the practical part. In the practical part are realized calculations. Counted performance of models is based on profit from market odds. In the period ex-post are calculated optimum model parameters that are used in the ex-ante period, where is calculated performance of the model. The thesis answers question: Are these models gaining from public database effective in modern age?

The development of distributed training strategies for statistical prediction functions is important for applications of machine learning, generally, and the development of distributed structured prediction training strategies is important for natural language processing (NLP), in particular. With ever-growing data sets this is, first, because, it is easier to increase computational capacity by adding more processor nodes than it is to increase the power of individual processor nodes, and, second, because data sets are often collected and stored in different locations. Iterative parameter mixing (IPM) is a distributed training strategy in which each node in a network of processors optimizes a regularized average loss objective on its own subset of the total available training data, making stochastic (per-example) updates to its own estimate of the optimal weight vector, and communicating with the other nodes by periodically averaging estimates of the optimal vector across the network. This algorithm has been contrasted with a close relative, called here the single-mixture optimization algorithm, in which each node stochastically optimizes an average loss objective on its own subset of the training data, operating in isolation until convergence, at which point the average of the independently created estimates is returned. Recent empirical results have suggested that this IPM strategy produces better models than the single-mixture algorithm, and the results of this thesis add to this picture. The contributions of this thesis are as follows. The first contribution is to produce and analyze an algorithm for decentralized stochastic optimization of regularized average loss objective functions. This algorithm, which we call the distributed regularized dual averaging algorithm, improves over prior work on distributed dual averaging by providing a simpler algorithm (used in the rest of the thesis), better convergence bounds for the case of regularized average loss functions, and certain technical results that are used in the sequel. The central contribution of this thesis is to give an optimization-theoretic justification for the IPM algorithm. While past work has focused primarily on its empirical test-time performance, we give a novel perspective on this algorithm by showing that, in the context of the distributed dual averaging algorithm, IPM constitutes a convergent optimization algorithm for arbitrary convex functions, while the single-mixture distribution algorithm is not. Experiments indeed confirm that the superior test-time performance of models trained using IPM, compared to single-mixture, correlates with better optimization of the objective value on the training set, a fact not previously reported. Furthermore, our analysis of general non-smooth functions justifies the use of distributed large-margin (support vector machine [SVM]) training of structured predictors, which we show yields better test performance than the IPM perceptron algorithm, the only version of the IPM to have previously been given a theoretical justification. Our results confirm that IPM training can reach the same level of test performance as a sequentially trained model and can reach better accuracies when one has a fixed budget of training time. Finally, we use the reduction in training time that distributed training allows to experiment with adding higher-order dependency features to a state-of-the-art phrase-structure parsing model. We demonstrate that adding these features improves out-of-domain parsing results of even the strongest phrase-structure parsing models, yielding a new state-of-the-art for the popular train-test pairs considered. In addition, we show that a feature-bagging strategy, in which component models are trained separately and later combined, is sometimes necessary to avoid feature under-training and get the best performance out of large feature sets.

One of the significant challenges facing industrial plant design is ensuring the integrity of massive design datasets generated during the project execution. This work is motivated by personal experience of data integrity issues during projects caused by insufficient automation affecting the quality of deliverables. Therefore, this project sought automated solutions for detecting anomalies in industrial design data in the form of outliers. Several novel methods are proposed, based on the Hidden Markov Model (HMM) and a modified General Regression Neural Network called the Margin-Based GRNN (MB-GRNN) along with optimisation techniques that minimise computation time. An HMM was used for validating process plant tag numbers using a self-learning approach. Results from experimental data show that HMM performance is equivalent to that of a custom-made design rule checking algorithm. The choice of components in industrial design involves setting specific design parameters that typically must reside inside permissible ranges called “design margins”. The MB-GRNN has the ability to estimate these permissible margins directly from design data and indicate potential design errors resulting from the invalid choice of design parameters by identifying data points outside of the estimated margins as outliers. The extremal permissible margin boundaries are determined by “stretching out” the upper and lower GRNN surfaces using an iterative application of stretch factors (a second kernel weighting factor). The method creates a variable insensitive band surrounding the data cloud, interlinked with the normal regression function, providing upper and lower margin boundaries. These boundaries can then be used to determine outliers and to predict a range of permissible values of design parameters during design. This method was compared to Parzen-Windows and another proximity-based method. The MB-GRNN also benefits from a modified algorithm for estimating the smoothing parameter using a combination of clustering, k-nearest neighbour, and localised covariance matrix. The computation time for this difficult task is minimised using a new derivative-based method that was tested successfully using a range of root-finding functions. The efficacy of the MB-GRNN and associated optimisation techniques were verified using three multivariate design datasets. The experimentation shows that the regression-based outlier classification approach used in this project complements the existing Parzen density-based method. These methods used in combination are intended for implementation as a decision support system for checking the quality of industrial design data to help minimise design and implementation costs. It is expected that the unsupervised techniques presented in this research work will benefit from the ever-increasing automation of industrial design processes.

Made available in DSpace on 2015-02-24T19:48:36Z (GMT). No. of bitstreams: 1 RobertoGO_Capa_ate_pag82.pdf: 4449318 bytes, checksum: a8a67621781bd4f76c4e2c0361bcc6a2 (MD5) Previous issue date: 2008-07-27
Conselho Nacional de Desenvolvimento Cient?fico e Tecnol?gico
The Borborema Province (BP) is a geologic domain located in Northeastern Brazil. The BP is limited at the south by the S?o Francisco craton, at the west by the Parna?ba basin, and both at the north and east by coastal sedimentary basins. Nonetheless the BP surface geology is well known, several key aspects of its evolution are still open, notably: i)its tectonic compartmentalization established after the Brasiliano orogenesis, ii) the architecture of its cretaceous continental margin, iii) the elastic properties of its lithosphere, and iv) the causes of magmatism and uplifting which occurred in the Cenozoic. In this thesis, a regional coverage of geophysical data (elevation, gravity, magnetic, geoid height, and surface wave global tomography) were integrated with surface geologic information aiming to attain a better understanding of the above questions. In the Riacho do Pontal belt and in the western sector of the Sergipano belt, the neoproterozoic suture of the collision of the Sul domain of the BP with the Sanfranciscana plate (SFP) is correlated with an expressive dipolar gravity anomaly. The positive lobule of this anomaly is due to the BP lower continental crust uplifting whilst the negative lobule is due to the supracrustal nappes overthrusting the SFP. In the eastern sector of the Sergipano belt, this dipolar gravity anomaly does not exist. However the suture still can be identified at the southern sector of the Maranc? complex arc, alongside of the Porto da Folha shear zone, where the SFP N-S geophysical alignments are truncated. The boundary associated to the collision of the Cear? domain of the BP with the West African craton is also correlated with a dipolar gravity anomaly. The positive lobule of this anomaly coincides with the Sobral-Pedro II shear zone whilst the negative lobule is associated with the Santa Quit?ria magmatic arc. Judging by their geophysical signatures, the major BP internal boundaries are: i)the western sector of the Pernambuco shear zone and the eastern continuation of this shear zone as the Congo shear zone, ii) the Patos shear zone, and iii) the Jaguaribe shear zone and its southwestern continuation as the Tatajuba shear zone. These boundaries divide the BP in five tectonic domains in the geophysical criteria: Sul, Transversal, Rio Grande do Norte, Cear?, and M?dio Corea?. The Sul domain is characterized by geophysical signatures associated with the BP and SFP collision. The fact that Congo shear zone is now proposed as part of the Transversal domain boundary implies an important change in the original definition of this domain. The Rio Grande do Norte domain presents a highly magnetized crust resulted from the superposition of precambrian and phanerozoic events. The Cear? domain is divided by the Senador Pompeu shear zone in two subdomains: the eastern one corresponds to the Or?s-Jaguaribe belt and the western one to the Cear?-Central subdomain. The latter subdomain exhibits a positive ENE-W SW gravity anomaly which was associated to a crustal discontinuity. This discontinuity would have acted as a rampart against to the N-S Brasiliano orogenic nappes. The M?dio Corea? domain also presents a dipolar gravity anomaly. Its positive lobule is due to granulitic rocks whereas the negative one is caused by supracrustal rocks. The boundary between M?dio Corea? and Cear? domains can be traced below the Parna?ba basin sediments by its geophysical signature. The joint analysis of free air anomalies, free air admittances, and effective elastic thickness estimates (Te) revealed that the Brazilian East and Equatorial continental margins have quite different elastic properties. In the first one 10 km < Te < 20 km whereas in the second one Te ? 10 km. The weakness of the Equatorial margin lithosphere was caused by the cenozoic magmatism. The BP continental margin presents segmentations; some of them have inheritance from precambrian structures and domains. The segmentations conform markedly with some sedimentary basin features which are below described from south to north. The limit between Sergipe and Alagoas subbasins coincides with the suture between BP and SFP. Te estimates indicates concordantly that in Sergipe subbasin Te is around 20 km while Alagoas subbasin has Te around 10 km, thus revealing that the lithosphere in the Sergipe subbasin has a greater rigidity than the lithosphere in the Alagoas subbasin. Additionally inside the crust beneath Sergipe subbasin occurs a very dense body (underplating or crustal heritage?) which is not present in the crust beneath Alagoas subbasin. The continental margin of the Pernambuco basin (15 < Te < 25 km) presents a very distinct free air edge effect displaying two anomalies. This fact indicates the existence in the Pernambuco plateau of a relatively thick crust. In the Para?ba basin the free air edge effect is quite uniform, Te ? 15 km, and the lower crust is abnormally dense probably due to its alteration by a magmatic underplating in the Cenozoic. The Potiguar basin segmentation in three parts was corroborated by the Te estimates: in the Potiguar rift Te ? 5 km, in the Aracati platform Te ? 25 km, and in the Touros platform Te ? 10 km. The observed weakness of the lithosphere in the Potiguar rift segment is due to the high heat flux while the relatively high strength of the lithosphere in the Touros platform may be due to the existence of an archaean crust. The Cear? basin, in the region of Munda? and Icara? subbasins, presents a quite uniform free air edge effect and Te ranges from 10 to 15 km. The analysis of the Bouguer admittance revealed that isostasy in BP can be explained with an isostatic model where combined surface and buried loadings are present. The estimated ratio of the buried loading relative to the surface loading is equal to 15. In addition, the lower crust in BP is abnormally dense. These affirmations are particularly adequate to the northern portion of BP where adherence of the observed data to the isostatic model is quite good. Using the same above described isostatic model to calculate the coherence function, it was obtained that a single Te estimate for the entire BP must be lower than 60 km; in addition, the BP north portion has Te around 20 km. Using the conventional elastic flexural model to isostasy, an inversion of crust thickness was performed. It was identified two regions in BP where the crust is thickened: one below the Borborema plateau (associated to an uplifting in the Cenozoic) and the other one in the Cear? domain beneath the Santa Quit?ria magmatic arc (a residue associated to the Brasiliano orogenesis). On the other hand, along the Cariri-Potiguar trend, the crust is thinned due to an aborted rifting in the Cretaceous. Based on the interpretation of free air anomalies, it was inferred the existence of a large magmatism in the oceanic crust surrounding the BP, in contrast with the incipient magmatism in the continent as shown by surface geology. In BP a quite important positive geoid anomaly exists. This anomaly is spatially correlated with the Borborema plateau and the Maca?-Queimadas volcanic lineament. The integrated interpretation of geoid height anomaly data, global shear velocity model, and geologic data allow to propose that and Edge Driven Convection (EDC) may have caused the Cenozoic magmatism. The EDC is an instability that presumably occurs at the boundary between thick stable lithosphere and oceanic thin lithosphere. In the BP lithosphere, the EDC mechanism would have dragged the cold lithospheric mantle into the hot asthenospheric mantle thus causing a positive density contrast that would have generated the main component of the geoid height anomaly. In addition, the compatibility of the gravity data with the isostatic model, where combined surface and buried loadings are present, together with the temporal correlation between the Cenozoic magmatism and the Borborema plateau uplifting allow to propose that this uplifting would have been caused by the buoyancy effect of a crustal root generated by a magmatic underplating in the Cenozoic
A Prov?ncia Borborema (PB) ? um dom?nio geol?gico-estrutural localizado no Nordeste do Brasil, limitado a sul pelo Cr?ton do S?o Francisco, a oeste pela Bacia do Parna?ba e a norte e leste pelas bacias costeiras. Embora bastante estudada por geologia de superf?cie, na PB ainda est?o em aberto aspectos importantes de sua evolu??o, notadamente: i) a sua compartimentagem tect?nica ap?s a Orog?nese Brasiliana, ii) a arquitetura da margem continental implantada no Cret?ceo, iii) as propriedades el?sticas de sua litosfera, e iv) as causas do magmatismo e do soerguimento no Cenoz?ico. Esta Tese empregou dados geof?sicos de cobertura regional (eleva??o, gravimetria, magnetometria, altura geoidal e tomografia), para aportar informa??es de geologia profunda aos problemas acima colocados. A sutura gerada pela colis?o neoproteroz?ica entre o Dom?nio Sul da PB e a Placa Sanfranciscana (PSF) ? marcada, na Faixa Riacho do Pontal e no oeste da Faixa Sergipana, por uma forte anomalia gravim?trica dipolar, cujo pico positivo corresponde ao al?amento da crosta inferior da PB e o negativo corresponde ?s nappes de supracrustais empurradas sobre a PSF. Na regi?o leste da Faixa Sergipana n?o h? assinaturas gravim?tricas que indiquem cavalgamento e flexura de placas, mas a interpreta??o de truncamentos de assinaturas geof?sicas de dire??o N-S da PSF permite localizar a sutura na margem sul do complexo de arco Maranc?, ao longo da Z. C. Porto da Folha. Por sua vez, o limite colisional do Dom?nio Cear? da PB com o Cr?ton Oeste-Africano, ao longo da Z. C. Sobral-Pedro II, ? tamb?m marcado por uma anomalia gravim?trica dipolar, cujo pico positivo coincide com a Z. C. Sobral-Pedro II, e o negativo coincide com o arco magm?tico de Santa Quit?ria. A julgar pela express?o geof?sica, os limites internos mais importantes da PB s?o: i) a Z. C. Pernambuco Oeste e sua continua??o na Z. C. Congo, ii) a Z. C. Patos e iii) a Z. C. Jaguaribe e sua continua??o na Z. C. Tatajuba. Estes limites dividem a PB em cinco grandes dom?nios geof?sicos-tect?nicos: Sul (ou Externo), Transversal, Rio Grande do Norte, Cear? e M?dio Corea?. O Dom?nio Sul ? marcado por assinaturas geof?sicas associadas ? colis?o da PB com a PSF. O Dom?nio Transversal teve a sua concep??o original de limites modificada porque a parte leste do seu limite sul foi associada com a Z. C. Congo. O Dom?nio Rio Grande do Norte apresenta a crosta mais magn?tica da PB, com superposi??o de fontes pr?-cambrianas e faneroz?icas. No Dom?nio Cear?, a Z. C. Senador Pompeu ? o divisor de dois subdom?nios: o leste corresponde ? Faixa Or?s-Jaguaribe e o oeste corresponde ao Cear?-Central, onde ocorre uma assinatura gravim?trica interpretada como uma descontinuidade crustal de dire??o ENE-WSW, que funcionou como um anteparo para as nappes brasilanas, com sentido de deslocamento para sul. O Dom?nio M?dio Corea? apresenta uma anomalia gravim?trica dipolar, cujo pico positivo est? associado com rochas granul?ticas, e o negativo com rochas supracrustais. A assinatura geof?sica do seu limite com o Dom?nio Cear? ? evidente, apesar dos sedimentos da Bacia do Parna?ba. A an?lise conjunta da anomalia ar-livre, admit?ncia ar-livre e estimativas da espessura el?stica efetiva (Te) evidenciou que as margens Leste e Equatorial possuem propriedades el?sticas bastante diferentes: enquanto a primeira tem Te entre 10 e 20 km, a segunda tem Te em torno ou inferior a 10 km. Essa diferen?a ? devida ao enfraquecimento da litosfera da Margem Equatorial produzida pelo magmatismo cenoz?ico. A margem continental da PB apresenta segmenta??es que incorporaram heran?as das estruturas e dos dom?nios pr?-cambrianos, que se correlacionam com os limites conhecidos das bacias. Descrevendo de sul para norte, o limite da separa??o da Bacia Sergipe - Alagoas em duas sub-bacias coincide com a sutura entre o Dom?nio Sul da PB e a PSF; as estimativas de Te indicam, concordantemente, que a Sub-bacia Sergipe (Te ? 20 km) se instalou em uma litosfera mais resistente do que a da Sub-bacia Alagoas (Te ? 10 km). Adicionalmente, no interior da crosta da Sub-bacia Sergipe ocorre um grande corpo denso (underplating ou heran?a crustal?) que n?o continua na Sub-bacia Alagoas. A margem da Bacia de Pernambuco (15 < Te < 25 km) apresenta caracter?sticas diferentes das outras bacias costeiras, porque no Plat? de Pernambuco h? duas anomalias do efeito de borda , o que indica a exist?ncia no plat? de uma crosta continental afinada, contudo ainda relativamente espessa. A Bacia da Para?ba se apresenta bastante uniforme, com Te em torno de 15 km, e possui uma crosta inferior relativamente densa, que foi interpretada como uma modifica??o por underplating magm?tico relacionado com o magmatismo cenoz?ico. A segmenta??o da Bacia Potiguar em tr?s partes ? corroborada pelas estimativas de Te: Rifte Potiguar (Te ? 5 km), Plataforma de Aracati (Te ? 25 km) e Plataforma de Touros (Te ? 10 km). A fragilidade da litosfera na regi?o do Rifte Potiguar est? associada com fluxo t?rmico atual alto, e a resist?ncia relativamente maior da Plataforma de Touros pode ser devida a uma crosta arqueana. A margem da Bacia do Cear?, no trecho das sub-bacias Munda? e Icara?, possui anomalia ar-livre uniforme, com Te entre 10 e 15 km. A an?lise da admit?ncia Bouguer revelou que a condi??o isost?tica da PB ? compat?vel com um modelo em que ocorrem carregamentos combinados na superf?cie e na base da crosta, com a carga da base 15 vezes maior que a do topo. Em adi??o, a PB possui uma crosta inferior anormalmente densa. Estas afirma??es s?o especialmente adequadas para a parte norte da PB porque a? a ader?ncia dos dados observados ao modelo ? maior. Para o mesmo modelo isost?tico e usando a fun??o coer?ncia, estimou-se que a Te da PB deve ser inferior a 60 km, embora sua por??o norte tenha Te de apenas 20 km. A invers?o de espessura de crosta, usando o modelo isost?tico com carga apenas na superf?cie, indicou que existem na PB duas regi?es de espessamento: uma abaixo do Planalto da Borborema (de origem cenoz?ica) e a outra no Dom?nio Cear?, sob o arco magm?tico de Santa Quit?ria (vestigial do Pr?-cambriano). Por outro lado, ocorre um afinamento ao longo do Trend Cariri-Potiguar, que representa o registro no interior do continente de um rifteamento cret?ceo abortado. A interpreta??o das anomalias ar-livre de fontes oce?nicas levou ? proposi??o de que ocorreu um volumoso magmatismo na ?rea oce?nica adjacente ? PB, ao contr?rio da ?rea continental, como indicam as informa??es de geologia de superf?cie. A PB apresenta uma expressiva anomalia positiva de ge?ide, com correla??o espacial com o Planalto da Borborema e o Alinhamento Macau-Queimadas. A integra??o de dados de tomografia de ondas superficiais e de anomalias residuais de ge?ide permitiu interpretar que uma convec??o em pequena escala (Edge Driven Convection-EDC), gerada na interface entre a raiz da litosfera continental fria e o manto quente da ?rea oce?nica, pode ter sido a causa do magmatismo cenoz?ico. O mecanismo de EDC teria causado o arrasto do manto litosf?rico continental frio para dentro do manto astenosf?rico quente, ocasionando assim contraste positivo de densidade, que seria uma componente importante da origem da anomalia de ge?ide. A compatibilidade dos dados gravim?tricos da PB com o modelo isost?tico que combina carregamentos no topo e na base da crosta, e a correla??o temporal entre o magmatismo cenoz?ico e o soerguimento do planalto, permite propor que o soerguimento deste ocorreu por causa do empuxo provocado pela raiz da crosta, produzida por um underplating magm?tico no Cenoz?ico

This thesis presents work around 3 themes: dark energy, gravitational waves and Bayesian inference. Both dark energy and gravitational wave physics are not yet well constrained. They present interesting challenges for Bayesian inference, which attempts to quantify our knowledge of the universe given our astrophysical data. A dark energy equation of state reconstruction analysis finds that the data favours the vacuum dark energy equation of state $w {=} -1$ model. Deviations from vacuum dark energy are shown to favour the super-negative ‘phantom’ dark energy regime of $w {< } -1$, but at low statistical significance. The constraining power of various datasets is quantified, finding that data constraints peak around redshift $z = 0.2$ due to baryonic acoustic oscillation and supernovae data constraints, whilst cosmic microwave background radiation and Lyman-$\alpha$ forest constraints are less significant. Specific models with a conformal time symmetry in the Friedmann equation and with an additional dark energy component are tested and shown to be competitive to the vacuum dark energy model by Bayesian model selection analysis: that they are not ruled out is believed to be largely due to poor data quality for deciding between existing models. Recent detections of gravitational waves by the LIGO collaboration enable the first gravitational wave tests of general relativity. An existing test in the literature is used and sped up significantly by a novel method developed in this thesis. The test computes posterior odds ratios, and the new method is shown to compute these accurately and efficiently. Compared to computing evidences, the method presented provides an approximate 100 times reduction in the number of likelihood calculations required to compute evidences at a given accuracy. Further testing may identify a significant advance in Bayesian model selection using nested sampling, as the method is completely general and straightforward to implement. We note that efficiency gains are not guaranteed and may be problem specific: further research is needed.

Today, the main research field for the automotive industry is to find solutions for active safety. In order to perceive the surrounding environment, tracking nearby traffic objects plays an important role. Validation of the tracking performance is often done in staged traffic scenarios, where additional sensors, mounted on the vehicles, are used to obtain their true positions and velocities. The difficulty of evaluating the tracking performance complicates its development. An alternative approach studied in this thesis, is to record sequences and use non-causal algorithms, such as smoothing, instead of filtering to estimate the true target states. With this method, validation data for online, causal, target tracking algorithms can be obtained for all traffic scenarios without the need of extra sensors. We investigate how non-causal algorithms affects the target tracking performance using multiple sensors and dynamic models of different complexity. This is done to evaluate real-time methods against estimates obtained from non-causal filtering. Two different measurement units, a monocular camera and a LIDAR sensor, and two dynamic models are evaluated and compared using both causal and non-causal methods. The system is tested in two single object scenarios where ground truth is available and in three multi object scenarios without ground truth. Results from the two single object scenarios shows that tracking using only a monocular camera performs poorly since it is unable to measure the distance to objects. Here, a complementary LIDAR sensor improves the tracking performance significantly. The dynamic models are shown to have a small impact on the tracking performance, while the non-causal application gives a distinct improvement when tracking objects at large distances. Since the sequence can be reversed, the non-causal estimates are propagated from more certain states when the target is closer to the ego vehicle. For multiple object tracking, we find that correct associations between measurements and tracks are crucial for improving the tracking performance with non-causal algorithms.

Approved for public release; distribution is unlimited
During the month of July 1987 an acoustical experiment was conducted by the United States Naval Research Laboratory (NRL) in the East Greenland Sea Marginal Ice Zone (MIZ) . Ambient noise "hot spots" or concentrated areas of relatively high noise levels were found along the ice edge using a towed array. Ambient noise levels were obtained on 27 and 28 July using AN/SSQ-57A and AN/SSQ-57XN5 calibrated sonobuoys . The temperature structure of the area was determined using XBT (ship) and AXBT (P3C aircraft) buoys placed inside and outside the ice edge. The ice edge was determined from coincident satellite photos, 90 GHz microwave imagery and P3 radar ice edge maps. Weather data (sea state and wind speed and direction) were recorded on the ship. The data seem to indicate a correlation between the high ambient noise levels of the hot spots and the presence of a large topographically controlled mesoscale eddy located at the southeastern extent of the MIZ.
http://archive.org/details/spatialvariabili00bigg
Lieutenant, United States Navy

Thesis (Ph. D.)--University of Wisconsin--Madison, 1988.
Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 99-103).

博士
國立交通大學
電子工程研究所
78
This dissertation presents some methods for finding the boundaries of constant gain margin and phase margin of control systems with adjustable parameters. The considered systems are first modified by adding a gain-phase margin tester, then the characteristic equations are formulated, and finally the stability equations are used to find the stability boundaries and the boundaries of constant gain margin and phase margin. The main contribution of the prsented methods is to obtain complete information about the effects of adjustable parameters on gain margins and phase margins and their corresponding crossover frequencies. In order to illustrate the usefullness of the presented methods, four types of linear control systems are considered: (1) continuous control system, (2) sampled-data control systems, (3) control systems with transport lags, (4) multiple-input multipleoutput control systems.

碩士
國立交通大學
電信工程系所
97
As the dimension of complementary metal-oxide-semiconductor (CMOS) devices shrunk into sub-65nm scale, the threshold voltage Vth fluctuation is pronounced and becomes crucial for the design window, yield, noise margin, stability, and reliability of ultra large-scale integration circuits. Various randomness effects resulted from the random nature of manufacturing process have induced significant fluctuations of electrical characteristics in nanometer scale (nanoscale) devices and circuits. In this thesis, a three-dimensional “atomistic” coupled device-circuit simulation is intensively performed to investigate the impact of intrinsic parameter fluctuations on 16-nm-gate planar metal-oxide-semiconductor field-effect-transistor (MOSFET) static random access memory (SRAM) cells. For device with 140 mV threshold voltage, the static noise margin (SNM) of 6T SRAM with unitary cell ratio is 20 mV with 80% normalized SNM fluctuation (σSNM), which may not ensure correct operation of circuits. Thus, improvement and suppression approaches based on the circuit and device viewpoints are implemented to examine the associated characteristics in 16-nm-gate SRAM cells. From the circuit viewpoint, an 8T planar SRAM architecture is explored. Compared with the conventional 6T SRAM, under the same Vth = 140 mV, the SNM is enlarged to 233 mV and the SNM is reduced to 22 mV (around 9.5% normalized SNM) at a cost of 30% extra chip area. To prevent the increase of chip area, silicon-on-insulator fin-type field-effect-transistors (SOI FinFETs) replaced the planar MOSFETs in 6T SRAM is further examined. The SNM of 6T SOI FinFETs SRAM is 125 mV and the normalized σSNM is suppressed significantly to 5.3% (6.8 mV in σSNM). The 8T SRAM architecture can provide largest SNM and is promising in near future design; however, to prevent the increase of chip area and suppress the intrinsic parameter fluctuations, development of fabrication for SOI FinFET SRAM is crucial for sub-22nm technology.

Documentos apresentados no âmbito do reconhecimento de graus e diplomas estrangeiros
After the development of Deepface and DeepID methods in 2014, deep learning methods for image recognition has dramatically improved the state-of-the-art performance on Deep Convolutional Neural Networks (DCNNs) and reshaped the research landscape of image processing and data analysis. In spite of rapid improvement in deep learning algorithms, it still has various challenges like adjustment of appropriate loss function and optimization strategy to handle large scale problems in many computer vision applications including Face Recognition (FR) and Handwritten Digit Recognition (HDR). This thesis focus on these challenges and their better solution.

In high range-resolution (HRR) radar systems, the returns from a single target may fall in multiple adjacent range bins which individually vary in amplitude. A target following this representation is commonly referred to as an extended target and results in more information about the target. However, extracting this information from the radar returns is challenging due to several complexities. These complexities include the single dimensional nature of the radar measurements, complexities associated with the scattering of electromagnetic waves, and complex environments in which radar systems are required to operate. There are several applications of HRR radar systems which extract target information with varying levels of success. A commonly used application is that of imaging referred to as synthetic aperture radar (SAR) and inverse SAR (ISAR) imaging. These techniques combine multiple single dimension measurements in order to obtain a single two dimensional image. These techniques rely on rotational motion between the target and the radar occurring during the collection of the single dimension measurements. In the case of ISAR, the radar is stationary while motion is induced by the target. There are several difficulties associated with the unknown motion of the target when standard Doppler processing techniques are used to synthesise ISAR images. In this dissertation, a non-standard Dop-pler approach, based on Bayesian inference techniques, was considered to address the difficulties. The target and observations were modelled with a non-linear state space model. Several different Bayesian techniques were implemented to infer the hidden states of the model, which coincide with the unknown characteristics of the target. A simulation platform was designed in order to analyse the performance of the implemented techniques. The implemented techniques were capable of successfully tracking a randomly generated target in a controlled environment. The influence of varying several parameters, related to the characteristics of the target and the implemented techniques, was explored. Finally, a comparison was made between standard Doppler processing and the Bayesian methods proposed.
Dissertation (MEng)--University of Pretoria, 2013.
gm2014
Electrical, Electronic and Computer Engineering
unrestricted