Neste trabalho, foi proposta uma nova família de distribuições, a qual permite modelar dados de sobrevivência quando a função de risco tem formas unimodal e U (banheira). Ainda, foram consideradas as modificações das distribuições Weibull, Fréchet, half-normal generalizada, log-logística e lognormal. Tomando dados não-censurados e censurados, considerou-se os estimadores de máxima verossimilhança para o modelo proposto, a fim de verificar a flexibilidade da nova família. Além disso, um modelo de regressão locação-escala foi utilizado para verificar a influência de covariáveis nos tempos de sobrevida. Adicionalmente, conduziu-se uma análise de resíduos baseada nos resíduos deviance modificada. Estudos de simulação, utilizando-se de diferentes atribuições dos parâmetros, porcentagens de censura e tamanhos amostrais, foram conduzidos com o objetivo de verificar a distribuição empírica dos resíduos tipo martingale e deviance modificada. Para detectar observações influentes, foram utilizadas medidas de influência local, que são medidas de diagnóstico baseadas em pequenas perturbações nos dados ou no modelo proposto. Podem ocorrer situações em que a suposição de independência entre os tempos de falha e censura não seja válida. Assim, outro objetivo desse trabalho é considerar o mecanismo de censura informativa, baseado na verossimilhança marginal, considerando a distribuição log-odd log-logística Weibull na modelagem. Por fim, as metodologias descritas são aplicadas a conjuntos de dados reais. / In this study, a new family of distributions was proposed, which allows to model survival data when the function of risk has unimodal shapes and U (bathtub). Modifications of the Weibull, Fréchet, generalized half-normal, log-logistic and lognormal distributions were considered. Taking censored and non-censored data, we consider the maximum likelihood estimators for the proposed model, in order to check the flexibility of the new family. Also, it was considered a location-scale regression model, to verify the influence of covariates on survival times. Additionally, a residual analysis was conducted based on modified deviance residuals. For different parameters fixed, percentages of censoring and sample sizes, several simulation studies were performed with the objective of verify the empirical distribution of the martingale type and modified deviance residuals. To detect influential observations, measures of local influence were used, which are diagnostic measures based on small perturbations in the data or in the proposed model. It can occur situations in which the assumption of independence between the failure and censoring times is not valid. Thus, another objective of this work is to consider the informative censoring mechanism based on the marginal likelihood, considering the log-odd log-logistic Weibull distribution in modelling. Finally, the methodologies described are applied to sets of real data.

Neste trabalho, foi proposta uma nova família de distribuições, a qual permite modelar dados de sobrevivência quando a função de risco tem formas unimodal e U (banheira). Ainda, foram consideradas as modificações das distribuições Weibull, Fréchet, half-normal generalizada, log-logística e lognormal. Tomando dados não-censurados e censurados, considerou-se os estimadores de máxima verossimilhança para o modelo proposto, a fim de verificar a flexibilidade da nova família. Além disso, um modelo de regressão locação-escala foi utilizado para verificar a influência de covariáveis nos tempos de sobrevida. Adicionalmente, conduziu-se uma análise de resíduos baseada nos resíduos deviance modificada. Estudos de simulação, utilizando-se de diferentes atribuições dos parâmetros, porcentagens de censura e tamanhos amostrais, foram conduzidos com o objetivo de verificar a distribuição empírica dos resíduos tipo martingale e deviance modificada. Para detectar observações influentes, foram utilizadas medidas de influência local, que são medidas de diagnóstico baseadas em pequenas perturbações nos dados ou no modelo proposto. Podem ocorrer situações em que a suposição de independência entre os tempos de falha e censura não seja válida. Assim, outro objetivo desse trabalho é considerar o mecanismo de censura informativa, baseado na verossimilhança marginal, considerando a distribuição log-odd log-logística Weibull na modelagem. Por fim, as metodologias descritas são aplicadas a conjuntos de dados reais. / In this study, a new family of distributions was proposed, which allows to model survival data when the function of risk has unimodal shapes and U (bathtub). Modifications of the Weibull, Fréchet, generalized half-normal, log-logistic and lognormal distributions were considered. Taking censored and non-censored data, we consider the maximum likelihood estimators for the proposed model, in order to check the flexibility of the new family. Also, it was considered a location-scale regression model, to verify the influence of covariates on survival times. Additionally, a residual analysis was conducted based on modified deviance residuals. For different parameters fixed, percentages of censoring and sample sizes, several simulation studies were performed with the objective of verify the empirical distribution of the martingale type and modified deviance residuals. To detect influential observations, measures of local influence were used, which are diagnostic measures based on small perturbations in the data or in the proposed model. It can occur situations in which the assumption of independence between the failure and censoring times is not valid. Thus, another objective of this work is to consider the informative censoring mechanism based on the marginal likelihood, considering the log-odd log-logistic Weibull distribution in modelling. Finally, the methodologies described are applied to sets of real data.

In this thesis, we aim at improving the search result quality by utilizing the search intelligence (history of searches) available in the form of click-through data. We address two key issues, namely 1) relevance feedback extraction and fusion, and 2) deciphering search query intentions. Relevance Feedback Extraction and Fusion: The existing search engines depend heavily on the web linkage structure in the form of hyperlinks to determine the relevance and importance of the documents. But these are collective judgments given by the page authors and hence, prone to collaborated spamming. To overcome the spamming attempts and language semantic issues, it is also important to incorporate the user feedback on the documents' relevance. Since users can be hardly motivated to give explicit/direct feedback on search quality, it becomes necessary to consider implicit feedback that can be collected from search engine logs. Though a number of implicit feedback measures have been proposed in the literature, we have not been able to identify studies that aggregate those feedbacks in a meaningful way to get a final ranking of documents. In this thesis, we first evaluate two implicit feedback measures namely 1) click sequence and 2) time spent on the document for their content uniqueness. We develop a mathematical programming model to collate the feedbacks collected from different sessions into a single ranking of documents. We use Kendall's τ rank correlation to determine the uniqueness of the information content present in the individual feedbacks. The experimental evaluation on top 30 select queries from an actual search log data confirms that these two measures are not in perfect agreement and hence, incremental information can potentially be derived from them. Next, we study the feedback fusion problem in which the user feedbacks from various sessions need to be combined meaningfully. Preference aggregation is a classical problem in economics and we study a variation of it where the rankers, i.e., the feedbacks, possess different expertise. We extend the generalized Mallows' model to model the feedback rankings given in user sessions. We propose a single stage and two stage aggregation framework to combine different feedbacks into one final ranking by taking their respective expertise into consideration. We show that the complexity of the parameter estimation problem is exponential in number of documents and queries. We develop two scalable heuristics namely, 1) a greedy algorithm, and 2) a weight based heuristic, that can closely approximate the solution. We also establish the goodness of fit of the model by testing it on actual log data through log-likelihood ratio test. As the independent evaluation of documents is not available, we conduct experiments on synthetic datasets devised appropriately to examine the various merits of the heuristics. The experimental results confirm the possibility of expertise oriented aggregation of feedbacks by producing orderings better than both the best ranker as well as equi-weight aggregator. Motivated with this result, we extend the aggregation framework to hold infinite rankings for the meta-search applications. The aggregation results on synthetic datasets are found to be ensuring the extension fruitful and scalable. Deciphering Search Query Intentions: The search engine often retrieves a huge list of documents based on their relevance scores for a given query. Such a presentation strategy may work if the submitted query is very specific, homogeneous and unambiguous. But many a times it so happen that the queries posed to the search engine are too short to be specific and hence ambiguous to identify clearly the exact information need, (eg. "jaguar"). These ambiguous and heterogeneous queries invite results from diverse topics. In such cases, the users may have to sift through the entire list to find their needed information and that could be a difficult task. Such a task can be simplified by organizing the search results under meaningful subtopics, which would help the users to directly move on to their topic of interest and ignore the rest. We develop a method to determine the various possible intentions of a given short generic and ambiguous query using information from the click-through data. We propose a two stage clustering framework to co-cluster the queries and documents into intentions that can readily be presented whenever it is demanded. For this problem, we adapt the spectral bipartite partitioning by extending it to automatically determine the number of clusters hidden in the log data. The algorithm has been tested on selected ambiguous queries and the results demonstrate the ability of the algorithm in distinguishing among the user intentions.

Thesis (MScEng (Electrical and Electronic Engineering))--University of Stellenbosch, 2010. / ENGLISH ABSTRACT: This paper presents the implementation of an artificial magnetic ground-plane with a low profile Log-periodic Dipole Array (LPDA) antennas. After the properties of three typical Electromagnetic Bandgap (EBG) structures are investigated and their bandwidth properties are studied, a mechanism is presented to improve the band-width over which the EBG surface acts as a perfect magnetic conductor (PMC). A low profile LPDA is modeled above this surface and the results indicate an improved band-width region. Compared with a LPDA in free space the frequency band is shifted higher by the EBG surface and the gain pattern is shifted from a horizontal orientation to a vertical orientation. / AFRIKAANSE OPSOMMING: Hierdie dokument stel voor die implementering van kunsmatige magnetiese grondvlakke met Logaritmiese Periodiese Dipool Samestelling (LPDS) antennas. Die eienskappe van drie tipiese Elektromagnetiese Bandgaping (EBG) strukture word ondersoek en hul bandwydte eienskappe word bestudeer. ’n Meganisme word voorgestel om die bandwydte te verbeter waar die EBG oppervlakte soos n perfekte magnetiese geleier optree. ’n Lae profiel LPDS word bo hierdie oppervlakte geplaas. Die resultate dui aan ’n verbetering in the bandwydte. In vergelyking met ’n LPDS in vrye ruimte skuif die frekwensie band ho¨er as gevolg van die EBG oppervlakte en die aanwins patroon skuif van ’n horisontale orientasie na ’n vertikale orientasie.

Title: Models of Marked Point Processes Author: Ivan Héda Department: Department of Probability and Mathematical Statistics Supervisor: doc. RNDr. Zbyněk Pawlas, Ph.D. Abstract: In the first part of the thesis, we present necessary theoretical basics as well as the definition of functional characteristics used for examination of marked point patterns. Second part is dedicated to review some known marking strategies. The core of the thesis lays in the study of intensity-marked point processes. General formula for the characteristics is proven for this marking strategy and general class of the models with analytically computable characteristics is introduced. This class generalizes some known models. Theoretical results are used for real data analysis in the last part of the thesis. Keywords: marked point process, marked log-Gaussian Cox process, intensity-marked point process 1

Survival analysis is an effective tool for conservation studies, since it measure the risk of an event that is important for the survival of populations and preservation of biodiversity. In this thesis three different models for survival analysis are used to estimate the age at breeding between two genetically different populations of brown trout. These populations are an evolutionary enigma, since they apparently coexist in direct competition with each other, which according to ecological theory should not happen. Thus it is of interest if differences between them can be identified. The data consists of brown trouts and has been collected over 20 years. The models are the Cox Proportional Hazards model, the Complementary Log-Log Link model and the Log Logistic Accelerated Failure-Time model. The Cox model were estimated in three different ways due to the nonproportional hazards in the estimates of time to breeding, which gave different interpretations of the same model. All of the models agree that the population B breed at younger ages than the population A, which suggests that the two populations have different reproductive strategies.

Neste trabalho é proposta uma nova distribuição de quatro parâmetros denominada distribuição log-logística exponenciada geométrica, baseada em um mecanismo de dupla ativação para modelar dados de tempo de vida. Para esta nova distribuição, foi realizado um estudo da função de densidade de probabilidade, da função de distribuição acumulada, da função de sobrevivência e da função de taxa de falha, a qual apresenta formas que podem modelar dados de tempo de vida, tais como: forma crescente, decrescente, unimodal, bimodal e forma de U. Obteve-se expansões da função de densidade, expressões para os momentos de probabilidade ponderada, função geradora de momentos, desvios médios e as curvas de Bonferroni e de Lorenz. Considerando dados censurados, foi utilizado o método de máxima verossimilhança para estimação dos parâmetros. Analogamente também é proposto um modelo de regressão baseado no logaritmo da distribuição log-logística exponenciada geométrica com dupla ativação, que é uma extensão dos modelos de regressão logística exponenciada e logística. Este modelo pode ser usado na análise de dados reais, por fornecer um melhor ajuste que os modelos de regressão particulares, logística exponenciada e logística. Finalmente, são apresentados duas aplicações para ilustrar a utilização da nova distribuição. / In this work, we propose a new distribution with four parameters the so called exponentiated log-logistic geometric distribution based on a double mechanism of activation for modeling lifetime data. For this new distribution, we study the density function, cumulative distribution, survival function and the failure rate function which allows major harzad rates: increasing, decreasing, bathtub, unimodal and bimodal failure rates. We also obtain the density function expansions and the expressions for the probability-weighted moments, moment generating function, mean deviation and Bonferroni and Lorenz curves. Considering censored data, we use the maximum likelihood method for estimating the parameters. Similarly, we also propose the regression model based on the logarithm of the exponentiated log-logistic geometric distribution with double activation, which is an extension of the exponential logistic and logistic regression models. This new model could be widely used in the analysis of real data to provide a better fit than exponetial logistic and logistic regression models. Finally, two applications are presented to illustrate the application of the new distribution.

Made available in DSpace on 2016-04-04T18:31:34Z (GMT). No. of bitstreams: 1 Carlos Alberto Tercariol Bergonso.pdf: 948199 bytes, checksum: 9baee66f05960fc08d06f925a9ea859a (MD5) Previous issue date: 2012-12-07 / The present dissertation proposes a 802.11 network evaluation sequence which can be adopted, independently of the equipment used at the network. The log-distance path loss model is used for the environment characterization (by the model ? parameter) where the network is evaluated. The specific objectives of each evaluation stage are presented and contextualized. The full proposal has initial steps related to the information determination which will be used as references of equipment location and data, and of network performance. Those reference data will be later compared to the handled data obtained from measurements related to the physical and MAC (Medium Access Control) layer performance and behavior, as well as to the transport layer. By the end, one stage to consolidate the measured data is necessary in order to determine a single final result, weigh up according to the specific objectives of the network and/or equipment evaluation. However, the work focus is at the equipment data obtaining stage and at the lower layers (physical and MAC) performance/behavior determination, so that evaluation methods are developed for those two stages. It is demonstrated that the effective transmission physical layer data rates vary much more than what is expected due to the interpretation of the data provided by equipment manufactors. And that better performance could be obtained from complex networks, which made use of multiple access points, installed at unfavorable propagation environments, depending upon the network project. / A presente disserta??o prop?e uma sequ?ncia de avalia??o para redes 802.11 que pode ser adotada, independentemente de quais s?o os equipamentos utilizados na rede. O modelo de log-distance path loss ? utilizado para caracteriza??o do ambiente (pelo par?metro ? do modelo) em que a rede ? avaliada. Os objetivos espec?ficos de cada uma das etapas de avalia??o s?o apresentadas e contextualizadas. A proposta, em sua totalidade, apresenta etapas iniciais relacionadas ? determina??o de informa??es que ser?o usadas como refer?ncias de localidade e desempenho de rede, e de dados dos equipamentos. Esses dados de refer?ncia ser?o posteriormente comparados com os dados coletados e tratados de medi??es realizadas em rela??o ao desempenho e comportamento das camadas f?sicas e MAC (Medium Access Control) e da camada de transporte. Por fim, uma etapa de consolida??o dos resultados das medi??es ? necess?ria para se obter um resultado final ?nico, composto de forma a ponderar de acordo com o(s) objetivo(s) espec?fico(s) de avalia??o da rede e/ou equipamento. O foco dessa disserta??o, contudo, est? na obten??o de dados dos equipamentos, de desempenho e comportamento das camadas inferiores (f?sica e MAC), de forma que para essas etapas, m?todos de avalia??o s?o desenvolvidos. ? demonstrado que as taxas de transmiss?o de dados efetivamente praticadas na camada f?sica variam muito mais que o esperado em fun??o da interpreta??o dos dados fornecidos pelos fabricantes de equipamentos. E que, dependendo do projeto da rede, pode-se obter melhores desempenhos em ambientes de propaga??o desfavor?vel que utilizam m?ltiplos pontos de acesso.

Cette thèse porte sur les plongements d'espaces homogènes sphériques sur un corps quelconque. Dans une première partie, on aborde la classification de ces plongements, dans la lignée des travaux de Demazure et bien d'autres sur les variétés toriques, et de Luna, Vust et Knop sur les variétés sphériques. Dans une seconde partie, on généralise en caractéristique positive certains résultats obtenus par Bien et Brion portant sur les plongements complets et lisses qui sont log homogènes, c'est-à-dire dont le bord est un diviseur à croisements normaux et le fibré tangent logarithmique associé est engendré par ses sections globales. Dans une dernière partie, on construit par éclatements successifs une compactification lisse et log homogène explicite du groupe linéaire (différente de celle obtenue par Kausz). En prenant dans cette compactification les points fixes de certains automorphismes, on en déduit alors la construction de compactifications lisses et log homogènes de certains groupes semi-simples classiques. / This thesis is devoted to the study of embeddings of spherical homogeneous spaces over an arbitrary field. In the first part, we address the classification of such embeddings, in the spirit of Demazure and many others in the setting of toric varieties and of Luna, Vust and Knop in the setting of spherical varieties. In the second part, we generalize in positive characteristics some results obtained by Bien and Brion on those complete smooth embeddings that are log homogeneous, i.e., whose boundary is a normal crossing divisor and the associated logarithmic tangent bundle is generated by its global sections. In the last part, we construct an explicit smooth log homogeneous compactification of the general linear group by successive blow-ups (different from the one obtained by Kausz). By taking fixed points of certain automorphisms on this compactification, one gets smooth log homogeneous compactifications of some classical semi-simple groups.

O padrão espacial de espécies em florestas nativas pode fornecer evidências sobre a estrutura da comunidade vegetal. Fatores ambientais podem influenciar o padrão espacial das espécies, como as características edáficas e processos que dependem da densidade, como competição intra e interespecífica. Desse modo, a pesquisa da relação entre as características ambientais e o padrão espacial de espécies florestais pode ajudar a entender a dinâmica de florestas. O objetivo deste estudo foi empregar técnicas da análise de processos pontuais para verificar o efeito de fatores ambientais sobre a ocorrência de espécies florestais. A área de estudo foi a Estação Ecológica de Assis (EEA), da unidade de Conservação do Estado de São Paulo em parcelas permanentes, dentro do projeto \"Diversidade, dinâmica e conservação em florestas do Estado de São Paulo: 40 ha de parcelas permanentes\" do programa Biota da FAPESP. A descrição do padrão espacial das espécies mais abundantes na área de estudo foi avaliada pela função K proposta por Ripley e suas extensões para processo não homogêneos, por meio das coordenadas geográficas das espécies com circunferência na altura do peito igual ou superior a 15 cm. Modelos do Processo Poisson Homogêneo, Processo Poisson Não Homogêneos e do Processo Log Gaussiano de Cox foram ajustados para cada espécie. Foi utilizado o critério de AIC para selecionar o modelo que melhor se ajusta aos dados. Testes de diagnósticos dos modelos foram feitos utilizando a função K não homogênea sob a hipótese de Completa Aleatoriedade Espacial. Os resultados indicaram que as espécies mais abundantes na EEA apresentam um padrão de distribuição agregado, ou seja, o número esperado de indivíduos próximos de um evento qualquer é maior do que esperado para uma distribuição aleatória. Conforme esperado, os fatores ambientais desempenharam um importante papel para explicar a distribuição espacial das espécies, porém, os resultados indicaram que existe uma variação espacialmente estruturada que não foi incluída na análise que é imprescindível para um bom ajuste dos modelos. Portanto os resultados sugerem que outros fatores não incluídos nos modelos e dados disponíveis podem estar determinando os padrões espaciais além das (co)variáveis medidas. / The spatial pattern of species in native forests may provide evidence on the structure of the plant community. Environmental factors may influence the species\' spatial patterns, as well as soil characteristics and processes which depend on the density, as intraspecific and interspecific competition. Therefore, researching the relationship among the environmental features and the spatial pattern of the forest species may aid in understanding forest dynamics. The goal of this study was to apply point process techniques to verify the effect of environmental factors on the occurence of forest species. The study area was the \"Assis\'s Ecological Station\" (AES), of the \"Unit of conservation of the state of São Paulo in permanent plots\". The data was collected as part of the project entitled \"Diversity, dynamics and conservation in forests of São Paulo state: 40 ha of permanent plots\", from FAPESP\'s Biota program. The description of the spatial pattern of the most abundant species in the study area was assessed using Ripley\'s K function, using the species\' geographic coordinates with circumference at chest height equal or larger than 15 cm. Homogeneous and Non-Homogeneous Poisson Process models, as well as Cox Log Gaussian Process models were fitted to each species. Model selection was made using the Akaike information criterion. Diagnostics tests were made using the non-homogeneous K function under the hypothesis of complete spatial randomness. Results suggested that the most abundant species in the AES present an aggregate distribution pattern, i.e., the expected number of individuals next to any event is larger than the expected by a random distribution. As it was expected, environmental factors played a major role in explaining the spatial distribution of the species. However, results suggested that there is a spatially structured variation that was not included in the analysis and is needed to a good model fit. Therefore, further studies are needed to assess which environmental feature which was not considered in this study presents an effect on the occurence of these forest species