<div>Classification is an important supervised learning technique with numerous applications. This dissertation addresses two research problems in this area. The first is multicategory classification methods for high dimensional data. To handle high dimension low sample size (HDLSS) data with uneven group sizes (i.e., imbalanced data), we develop a new classification method called angle-based multicategory distance-weighted support vector machine (MDWSVM). It is motivated from its binary counterpart and has the merits of both the support vector machine (SVM) and distance-weighted discrimination (DWD) methods while alleviating both the data piling issue of SVM and the imbalanced data issue of DWD. Theoretical results and numerical studies are used to demonstrate the advantages of our MDWSVM method over existing methods.</div><div><br></div><div>The second part of the dissertation is on the application of classification methods to precision medicine problems. Because one-stage precision medicine problems can be reformulated as weighted classification problems, the subtle differences between classification methods may lead to different application performances under this setting. Among the margin-based classification methods, we propose to use the distance weighted discrimination outcome weighted learning (DWD-OWL) method. We also extend the model to handle negative rewards for better generality and apply the angle-based idea to handle multiple treatments. The proofs of Fisher consistency for DWD-OWL in both the binary and multicategory cases are provided. Under mild conditions, the insensitivity of DWD-OWL for imbalanced setting is also demonstrated.</div>

O vírus da hepatite E (HEV) é um patógeno emergente de distribuição global, causador de hepatite aguda e crônica em humanos e infecções assintomáticas em animais. No Brasil a prevalência de infecção por HEV em humanos e animais ainda é pouco compreendida, assim como as características de virulência, patogenicidade e de infecção inter-espécies de isolados do genótipo 3, zoonótico, circulantes no país também são desconhecidas. O estudo foi dividido em duas etapas, com os objetivos de 1) contribuir no diagnóstico laboratorial molecular do HEV a partir do desenvolvimento de um protocolo de RT-PCR em tempo real (RT-qPCR) para pesquisa do HEV em amostras biológicas, e 2) contribuir com a compreensão das características moleculares, sorológicas, clínico-laboratoriais, ultrassonográficas e histopatológicas associadas à infecção experimental em macacos-da-noite (Aotus azarai infulatus) por um isolado do genótipo 3 suíno do HEV previamente detectado na Amazônia oriental brasileira. O protocolo de RT-qPCR foi desenvolvido com a caracterização da curva de detecção e aplicado em concomitância com testes sorológicos para avaliação diagnóstica restrospectiva de 318 (n = 318) amostras de soros humanos suspeitos de hepatite E. O HEV-RNA não foi detectado em nenhuma das amostras humanas testadas, mas foi determinada soroprevalência de 3,4% e 5,9% de anti-HEV IgM e IgG, respectivamente, o que indicou baixa prevalência de infecção por HEV, mesmo entre pacientes com suspeita clínica e/ou laboratorial de hepatite E na Amazônia brasileira. O estudo experimental em macacos-da-noite foi desenvolvido durante 12 semanas e os animais infectados, por via intravenosa (n=3) e oral (n=3) (e dois controles), foram avaliados semanalmente para determinação dos parâmetros clínicos, bioquímicos, hematológicos, sorológicos (pesquisa de anti-HEV IgM e IgG por enzimaimunoensaio) e moleculares (HEV-RNA soro e fezes por RT-qPCR). Adicionalmente, os animais também foram submetidos a avaliação hepática mensal por ultrassonografia, histopatologia e pesquisa hepática de antígenos do HEV por imunohistoquímica. Os seis macacos-da-noite infectados apresentaram o HEV-RNA em amostras de soro e/ou fezes, e alguns apresentaram evidências de soroconversão, detecção hepática do antígeno viral por imunohistoquímica associada a alterações clínicas e laboratoriais de hepatite aguda oligossintomática. Assim, o protocolo RT-qPCR demonstrou ser aplicável na pesquisa molecular do HEV em amostras de humanos e animais, representando uma importante ferramenta de diagnóstico laboratorial. O estudo experimental permitiu a primeira validação de um primata neotropical como modelo experimental para estudos de infecção com o genótipo 3 do HEV. / Hepatitis E virus (HEV) is an emerging pathogen with global distribution that causes acute and chronic hepatitis in humans and asymptomatic infections in animals. In Brazil, the prevalence of HEV infection in humans and animals is still poorly understood, and the characteristics of virulence, pathogenicity and inter-species infection of the genotype 3 isolates circulating in the country are unknown. The study was divided in two stages that aimed to 1) contribute to the molecular diagnosis of HEV infection by the development of a real-time RT-PCR protocol (RT-qPCR) for HEV-RNA research in biological samples, and 2) to contribute to understanding of molecular, serological, clinical-laboratory, ultrasonographic and histopathological features of HEV genotype 3 in owl monkeys (Aotus azari infulatus) experimental infected with isolate of swine HEV genotype 3 previously detected in the Eastern Brazilian Amazon. The RT-qPCR protocol was developed with characterization of a quantification standard curve and later applied concurrently with serological tests in the retrospective evaluation of 318 (n = 318) human serum samples of hepatitis E suspected cases. HEV-RNA was not detected in any of human tested samples, but seroprevalence of 3.4% and 5.9% was determined for anti-HEV IgM and IgG, respectively, that indicated a low prevalence of HEV infection, even among patients with clinical and/or laboratory suspicion of hepatitis E in the Brazilian Amazon. The experimental study in owl monkeys was developed during12 weeks and the animals were infected by intravenous (n = 3) and oral (n = 3) routes (and two negative controls) were evaluated for determination of clinical, biochemical, hematological, serological (anti-HEV IgM and IgG by enzyme immunoassay) and molecular (HEV-RNA serum and stool by RT-qPCR) parameters weekly. Additionally, the animals were also evaluated by hepatic ultrasonography, histopathology and immunohistochemistry research of HEV antigens in liver monthly. The six infected owl monkeys presentend HEV-RNA in serum and/or stool, and some monkeys presented with evidence of seroconversion, liver detection of HEV antigens by immunohistochemistry associated with clinical and/or laboratory findings of oligosymptomatic acute hepatitis. Thus, the RT-qPCR protocol demonstrated to be applicable in the molecular investigation of HEV infection in human and animal samples, and it also represented an important laboratory diagnostic tool. The experimental study allowed the validation of the first neotropical primate model for HEV genotype 3 infection studies.

Description logics (DLs) are knowledge representation formalisms with numerous applications and well-understood model-theoretic semantics and computational properties. SRIQ is a DL that provides the logical underpinning for the semantic web language OWL 2, which is the W3C standard for knowledge representation on the web. A central component of most DL applications is an efficient and scalable reasoner, which provides services such as consistency testing and classification. Despite major advances in DL reasoning algorithms over the last decade, however, ontologies are still encountered in practice that cannot be handled by existing DL reasoners. Consequence-based calculi are a family of reasoning techniques for DLs. Such calculi have proved very effective in practice and enjoy a number of desirable theoretical properties. Up to now, however, they were proposed for either Horn DLs (which do not support disjunctive reasoning), or for DLs without cardinality constraints. In this thesis we present a novel consequence-based algorithm for TBox reasoning in SRIQ - a DL that supports both disjunctions and cardinality constraints. Combining the two features is non-trivial since the intermediate consequences that need to be derived during reasoning cannot be captured using DLs themselves. Furthermore, cardinality constraints require reasoning over equality, which we handle using the framework of ordered paramodulation - a state-of-the-art method for equational theorem proving. We thus obtain a calculus that can handle an expressive DL, while still enjoying all the favourable properties of existing consequence-based algorithms, namely optimal worst-case complexity, one-pass classification, and pay-as-you-go behaviour. To evaluate the practicability of our calculus, we implemented it in Sequoia - a new DL reasoning system. Empirical results show substantial robustness improvements over well-established algorithms and implementations, and performance competitive with closely related work.

O vírus da hepatite E (HEV) é um patógeno emergente de distribuição global, causador de hepatite aguda e crônica em humanos e infecções assintomáticas em animais. No Brasil a prevalência de infecção por HEV em humanos e animais ainda é pouco compreendida, assim como as características de virulência, patogenicidade e de infecção inter-espécies de isolados do genótipo 3, zoonótico, circulantes no país também são desconhecidas. O estudo foi dividido em duas etapas, com os objetivos de 1) contribuir no diagnóstico laboratorial molecular do HEV a partir do desenvolvimento de um protocolo de RT-PCR em tempo real (RT-qPCR) para pesquisa do HEV em amostras biológicas, e 2) contribuir com a compreensão das características moleculares, sorológicas, clínico-laboratoriais, ultrassonográficas e histopatológicas associadas à infecção experimental em macacos-da-noite (Aotus azarai infulatus) por um isolado do genótipo 3 suíno do HEV previamente detectado na Amazônia oriental brasileira. O protocolo de RT-qPCR foi desenvolvido com a caracterização da curva de detecção e aplicado em concomitância com testes sorológicos para avaliação diagnóstica restrospectiva de 318 (n = 318) amostras de soros humanos suspeitos de hepatite E. O HEV-RNA não foi detectado em nenhuma das amostras humanas testadas, mas foi determinada soroprevalência de 3,4% e 5,9% de anti-HEV IgM e IgG, respectivamente, o que indicou baixa prevalência de infecção por HEV, mesmo entre pacientes com suspeita clínica e/ou laboratorial de hepatite E na Amazônia brasileira. O estudo experimental em macacos-da-noite foi desenvolvido durante 12 semanas e os animais infectados, por via intravenosa (n=3) e oral (n=3) (e dois controles), foram avaliados semanalmente para determinação dos parâmetros clínicos, bioquímicos, hematológicos, sorológicos (pesquisa de anti-HEV IgM e IgG por enzimaimunoensaio) e moleculares (HEV-RNA soro e fezes por RT-qPCR). Adicionalmente, os animais também foram submetidos a avaliação hepática mensal por ultrassonografia, histopatologia e pesquisa hepática de antígenos do HEV por imunohistoquímica. Os seis macacos-da-noite infectados apresentaram o HEV-RNA em amostras de soro e/ou fezes, e alguns apresentaram evidências de soroconversão, detecção hepática do antígeno viral por imunohistoquímica associada a alterações clínicas e laboratoriais de hepatite aguda oligossintomática. Assim, o protocolo RT-qPCR demonstrou ser aplicável na pesquisa molecular do HEV em amostras de humanos e animais, representando uma importante ferramenta de diagnóstico laboratorial. O estudo experimental permitiu a primeira validação de um primata neotropical como modelo experimental para estudos de infecção com o genótipo 3 do HEV. / Hepatitis E virus (HEV) is an emerging pathogen with global distribution that causes acute and chronic hepatitis in humans and asymptomatic infections in animals. In Brazil, the prevalence of HEV infection in humans and animals is still poorly understood, and the characteristics of virulence, pathogenicity and inter-species infection of the genotype 3 isolates circulating in the country are unknown. The study was divided in two stages that aimed to 1) contribute to the molecular diagnosis of HEV infection by the development of a real-time RT-PCR protocol (RT-qPCR) for HEV-RNA research in biological samples, and 2) to contribute to understanding of molecular, serological, clinical-laboratory, ultrasonographic and histopathological features of HEV genotype 3 in owl monkeys (Aotus azari infulatus) experimental infected with isolate of swine HEV genotype 3 previously detected in the Eastern Brazilian Amazon. The RT-qPCR protocol was developed with characterization of a quantification standard curve and later applied concurrently with serological tests in the retrospective evaluation of 318 (n = 318) human serum samples of hepatitis E suspected cases. HEV-RNA was not detected in any of human tested samples, but seroprevalence of 3.4% and 5.9% was determined for anti-HEV IgM and IgG, respectively, that indicated a low prevalence of HEV infection, even among patients with clinical and/or laboratory suspicion of hepatitis E in the Brazilian Amazon. The experimental study in owl monkeys was developed during12 weeks and the animals were infected by intravenous (n = 3) and oral (n = 3) routes (and two negative controls) were evaluated for determination of clinical, biochemical, hematological, serological (anti-HEV IgM and IgG by enzyme immunoassay) and molecular (HEV-RNA serum and stool by RT-qPCR) parameters weekly. Additionally, the animals were also evaluated by hepatic ultrasonography, histopathology and immunohistochemistry research of HEV antigens in liver monthly. The six infected owl monkeys presentend HEV-RNA in serum and/or stool, and some monkeys presented with evidence of seroconversion, liver detection of HEV antigens by immunohistochemistry associated with clinical and/or laboratory findings of oligosymptomatic acute hepatitis. Thus, the RT-qPCR protocol demonstrated to be applicable in the molecular investigation of HEV infection in human and animal samples, and it also represented an important laboratory diagnostic tool. The experimental study allowed the validation of the first neotropical primate model for HEV genotype 3 infection studies.

The purpose of this study was to examine the hunting strategies of birds of prey in a probability learning situation. One great horned owl and one red-tailed hawk served as subjects. Three boxes and associated perches, each representing different potential prey areas, were placed in a room adjacent to the birds' regular housing and served as test apparatus. One box were required to land on a perch in order to gain access to the associated box and to the potential prey. A discrete-trial, self-correction procedure was used. In Experiment I, Box 1 was loaded with a live laboratory mouse on 60% of the trials, Box 2 on 30% of the trials, and Box 3 on the remaining 10%. A response requirement of sitting on the perch 5 seconds was programmed. In experiment II, the probability of reinforcement was reversed for the 10 and 60% boxes and the response requirement was increased from 5 to 20 seconds. Experiment III returned the probability of reinforcement for each box to the values used in Experiment I. In Experiment III the mice were euthanized prior to each trial. The owl matched responses to probability of reinforcement in all three experiments while the hawk matched in Experiment II and showed matching toward two of the three boxes in Experiment III. In Experiment I the hawk had a Box 2 preference. This research extends the generality of the matching concept and suggests that predatory birds do not randomly hunt or hunt in only one location, but rather tend to search according to the probability of reinforcement for that location.

The Forest Inventory and Analysis (FIA), the United States’ forest census, measured sixty-six Mexican spotted owl nest stands in order gain insight into the structure and composition of the nest habitat of this threatened species. I used these data, along with the greater FIA database and the Forest Vegetation Simulator to explore questions surrounding the management of Mexican spotted owl habitat, specifically how to balance the objective of sustaining and enhancing nest habitat in face of increasing forest fire size and severity in the Southwest. My research consisted of three studies. The first study quantified the structure and composition of the Mexican spotted owl nest stands and scrutinized current evaluation criteria of nest habitat. The second study estimated how much of the Southwestern landscape is at risk to high-severity crown fire and how much of the landscape is suitable Mexican spotted owl nest habitat. The third study modeled forest dynamics and silvicultural intervention in potential Mexican spotted owl nest stands. The purpose of this research is to assist in management of Southwestern forests in order to decrease fire size and severity while sustaining and enhancing current and future Mexican spotted owl nest habitat.

In this research I investigated ecological attributes of Powerful Owls (Ninox strenua) in a continuum of habitats throughout the Yarra Valley corridor of Victoria, Australia. These habitats ranged from a highly urbanized parkland (the Yarra Valley Metropolitan Park) to a relatively undisturbed closed forest (Toolangi State Forest). Different aspects of the owls' ecology were investigated at six sites to determine whether their behaviour changed when they occupied habitats with different levels of urbanization and disturbance. The ecological attributes investigated were habitat utilization and habitat requirements (for both roosting and nesting), adult behaviour (through radio-tracking), juvenile behaviour and dispersal (through radio tracking), diet (through analysing regurgitated food pellets) and breeding success rates. A number of methods were used to capture adult Powerful Owls. These are described and their effectiveness discussed. The types of radio-transmitters and colour bands used for identification of owls are also described. The results showed that Powerful Owls are present and successfully breed in urban and suburban areas and that they can tolerate moderate levels of disturbance. However, Powerful Owls do require sites with high prey densities, roost trees and trees with suitable breeding hollows. In comparison with Powerful Owls living elsewhere in forests, the urban owls displayed higher tolerance levels to disturbance and were less selective in terms of habitat usage and diet. Home range sizes of urban Powerful Owls also appeared much smaller than those of the forest-dwelling Powerful Owls. This is probably due to the high prey densities in the urban areas. The ecology of the Powerful Owl is compared with that of two owl species from North America, the Northern Spotted Owl (Strix occidentalis caurind) and the Great Horned Owl (Bubo virginianus). In particular, I compared the similarities and differences in habitat requirements and breeding successes in different habitats for the three species. Overall, it would appear that urban areas can support Powerful Owls providing some old-growth trees are maintained to provide nest hollows. Implications for the long-term management of Powerful Owls in urban areas are also discussed.

El sentido filosófico del término ontología hace referencia a la esencia misma del ser, a su existencia (onto=ser). Para un sistema inteligente lo que existe es lo que puede representarse. En informática una ontología se ha definido como una conceptualización formal y compartida de un dominio. Actualmente las ontologías tienen un rol clave en el desarrollo de laWeb Semántica, pero también están siendo usadas en otros dominios como la biomedicina, agricultura, defensa, robótica y astronomía.En la investigación llevada a cabo en esta tesis nos hemos centrado en el diseño y desarrollo de métodos generales y técnicas basadas en la lógica para dar soporte y respaldar el desarrollo de ontologías en varias de sus fases. La biomedicina ha sido seleccionada como dominio de aplicación debido a la gran cantidad de recursos de conocimiento que hay disponibles. Sin embargo, nuestra tecnología puede ser extendida y aplicada en otros dominios.Cabe destacar que nuestros métodos y técnicas han sido diseñados para el lenguaje de definición de ontologías OWL (Ontology Web Language) y su lógica de descripción subyacente SROIQ; sin embargo, también pueden ser aplicados sobre otros formalismos de definición de ontologías (ej: OBO), siempre y cuando puedan ser expresados en OWL.Como prueba de concepto hemos implementado un conjunto de herramientas basadas en los métodos y técnicas diseñados en la tesis. Estas herramientas han sido desarrolladas sobre el interfaz OWL API y diseñadas para el editor de ontologías Protégé.No obstante podrían ser extendidas e integradas en otros marcos de trabajo como el toolkit de NeOn.

The aim of this thesis is the development of an infrastructure which is used for semantic retrieval of multimedia contents. Motivated by the needs of semantic search and retrieval of multimedia contents, operating directly on the MPEG-7 based annotations can be thought as a reasonable way for meeting these needs as MPEG-7 is a common standard providing a wide multimedia content description schema. However, it is clear that the MPEG-7 formalism is deficient about the semantics and reasoning support. From this perspective, additionally, we need to represent MPEG-7 descriptions in a new formalism in order to fill the gap about semantics and reasoning. Then, the semantic web and multimedia technologies intercept at this point of multimedia semantics. In this thesis, OWL Web Ontology Language, which is based on description logic has been utilized to model a connection between the ontology semantics and video metadata. Modeling the domain of the videos using ontologies and the MPEG-7 descriptions, and reasoning on the videos by the help of the logical formalism of these ontologies are the main objectives of the thesis.

The Unified Medical Language System Semantic Network (UMLS SN) being an upper-level abstraction of the biomedical domain has a complex structure due to many relationships, making it difficult for human orientation. Therefore, while the SN is a valuable source for modeling contents of the biomedical domain its usage is limited. NetON was designed and built for the automatic transformation of UMLS SN to OWL sublanguages to support semantic operations between biomedical systems. NetON uses advances in the Semantic Web, a candidate technology for sustaining knowledge intensive tasks. Ontology Web Language (OWL) sublanguage rules are used to represent information in UMLS SN. The major contribution of NetON is the opportunity of automatic transformation of UMLS SN to OWL sublanguages named as OWL Basic Species. The aim of NetON is maximum possible information transformation from UMLS SN. The only information that is not able to be transformed to any OWL Basic Species due to the lack of appropriate constructors in OWL standard is inheritance blockings in UMLS SN. In UMLS SN, there are unseen assertions that can be inferred by using inference rules on explicitly specified assertions which are not essentially valid for all the descendants. Deduction outcomes of any OWL reasoners on NetON OWL Basic Species will also include false positives due to the lack of inheritance blocking information. The algorithms of the second dimension consider the inheritance blocking information while executing inference rules. As this cannot be done by any OWL reasoner, the second dimension offers a solution for application developers.