A seleção de variáveis é uma importante fase para a construção de um modelo parcimonioso. Entretanto, as técnicas mais populares de seleção de variáveis, como, por exemplo, a seleção do melhor subconjunto de variáveis e o método stepwise, ignoram erros estocásticos inerentes à fase de seleção das variáveis. Neste trabalho, foram estudados procedimentos alternativos aos métodos mais populares para o modelo de riscos proporcionais de Cox e o modelo de Cox com fragilidade gama. Os métodos alternativos são baseados em verossimilhançaa penalizada e diferem dos métodos usuais de seleção de variáveis, pois têm como objetivo excluir do modelo variáveis não significantes estimando seus coeficientes como zero. O estimador resultante possui propriedades desejáveis com escolhas apropriadas de funções de penalidade e do parâmetro de suavização. A avaliação desses métodos foi realizada por meio de simulação e uma aplicação a um conjunto de dados reais foi considerada. / Variable selection is an important step when setting a parsimonious model. However, the most popular variable selection techniques, such as the best subset variable selection and the stepwise method, do not take into account inherent stochastic errors in the variable selection step. This work presents a study of alternative procedures to more popular methods for the Cox proportional hazards model and the frailty model. The alternative methods are based on penalized likelihood and differ from the usual variable selection methods, since their objective is to exclude from the model non significant variables, estimating their coefficient as zero. The resulting estimator has nice properties with appropriate choices of penalty functions and the tuning parameter. The assessment of these methods was studied through simulations, and an application to a real data set was considered.

This thesis focuses on two fundamental topics, specifically in medical statistics: the modelling of correlated survival datasets and the variable selection of the significant covariates and random effects. In particular, two types of survival data are considered: the classical survival datasets, where subjects are likely to experience only one type of event and the competing risks datasets, where subjects are likely to experience one of several types of event. In Chapter 2, among other topics, we highlight the importance of adding frailty terms on the proposed models in order to account for the association between the survival time and characteristics of subjects/groups. The main novelty of this thesis is to simultaneously select fixed effects and frailty terms through the proposed statistical models for each survival dataset. Chapter 3 covers the analysis of the classical survival dataset through the proposed Cox Proportional Hazard (PH) model. Utilizing a Cox PH frailty model, may increase the dimension of variable components and estimation of the unknown coefficients becomes very challenging. The method proposed for the analysis of classical survival datasets involves simultaneous variable selection on both fixed effects and frailty terms through penalty functions. The benefit of penalty functions is that they identify the non-significant parameters and set them to have a zero effect in the model. Hence, the idea is to 'doubly-penalize' the partial likelihood of the Cox PH frailty model; one penalty for each term. Estimation and selection implemented through Newton-Raphson algorithms, whereas closed iterative forms for the estimation and selection of fixed effects and prediction of frailty terms were obtained. For the selection of frailty terms, penalties imposed on their variances since frailties are random effects. Based on the same idea, we further extend the simultaneous variable selection in the competing risks datasets in Chapter 4, using extended cause-specific frailty models. Two different scenarios are considered for frailty terms; in the first case we consider that frailty terms vary among different types of events (similar to the fixed effects) whereas in the second case we consider shared frailties over all the types of events. Moreover, our 'individual penalization' approach allows for one covariate to be significant for some types of events, in contrast to the frequently used 'group-penalization' where a covariate is entirely removed when it is not significant over all the events. For both proposed methods, simulation studies were conduced and showed that the proposed procedure followed for each analysis works well in simultaneously selecting and estimating significant fixed effects and frailty terms. The proposed methods are also applied to real datasets analysis; Kidney catheter infections, Diabetes Type 2 and Breast Cancer datasets. Association of the survival times and unmeasured characteristics of the subjects was studied as well as a variable selection for fixed effects and frailties implemented successfully.

A seleção de variáveis é uma importante fase para a construção de um modelo parcimonioso. Entretanto, as técnicas mais populares de seleção de variáveis, como, por exemplo, a seleção do melhor subconjunto de variáveis e o método stepwise, ignoram erros estocásticos inerentes à fase de seleção das variáveis. Neste trabalho, foram estudados procedimentos alternativos aos métodos mais populares para o modelo de riscos proporcionais de Cox e o modelo de Cox com fragilidade gama. Os métodos alternativos são baseados em verossimilhançaa penalizada e diferem dos métodos usuais de seleção de variáveis, pois têm como objetivo excluir do modelo variáveis não significantes estimando seus coeficientes como zero. O estimador resultante possui propriedades desejáveis com escolhas apropriadas de funções de penalidade e do parâmetro de suavização. A avaliação desses métodos foi realizada por meio de simulação e uma aplicação a um conjunto de dados reais foi considerada. / Variable selection is an important step when setting a parsimonious model. However, the most popular variable selection techniques, such as the best subset variable selection and the stepwise method, do not take into account inherent stochastic errors in the variable selection step. This work presents a study of alternative procedures to more popular methods for the Cox proportional hazards model and the frailty model. The alternative methods are based on penalized likelihood and differ from the usual variable selection methods, since their objective is to exclude from the model non significant variables, estimating their coefficient as zero. The resulting estimator has nice properties with appropriate choices of penalty functions and the tuning parameter. The assessment of these methods was studied through simulations, and an application to a real data set was considered.

<p>I detta arbete undersöks hur bestraffningsmetoder för att bestraffa storleken på GP-program påverkar generaliseringsförmågan. Arbetet grundar sig på ett arbete som Cavaretta och Chellapilla gjort, där de undersöker skillnaden i generaliseringsförmåga mellan bestraffningsmetoden ”Complexity Penalty functions” och ingen bestraffningsmetod.</p><p>I detta arbete har nya experiment gjorts med ”Complexity Penalty functions” och ”Adaptive parsimony pressure”, som är en annan bestraffningsmetod. Dessa bestraffningsmetoder har undersökts i samma domän som Cavaretta och Chellapilla och ytterligare i en domän för att ge en bättre bild av hur de generaliserar.</p><p>I arbetet visar det sig att användningen av någon av bestraffningsmetoderna ”Complexity Penalty functions” och ”Adaptive parsimony pressure” oftast ger bättre generaliseringsförmåga hos GP-program. Detta motsäger det Cavaretta och Chellapilla kommer fram till i sitt arbete. ”Adaptive parsimony pressure” verkar också vara bättre på att generalisera än ”Complexity Penalty functions”.</p>

I detta arbete undersöks hur bestraffningsmetoder för att bestraffa storleken på GP-program påverkar generaliseringsförmågan. Arbetet grundar sig på ett arbete som Cavaretta och Chellapilla gjort, där de undersöker skillnaden i generaliseringsförmåga mellan bestraffningsmetoden ”Complexity Penalty functions” och ingen bestraffningsmetod. I detta arbete har nya experiment gjorts med ”Complexity Penalty functions” och ”Adaptive parsimony pressure”, som är en annan bestraffningsmetod. Dessa bestraffningsmetoder har undersökts i samma domän som Cavaretta och Chellapilla och ytterligare i en domän för att ge en bättre bild av hur de generaliserar. I arbetet visar det sig att användningen av någon av bestraffningsmetoderna ”Complexity Penalty functions” och ”Adaptive parsimony pressure” oftast ger bättre generaliseringsförmåga hos GP-program. Detta motsäger det Cavaretta och Chellapilla kommer fram till i sitt arbete. ”Adaptive parsimony pressure” verkar också vara bättre på att generalisera än ”Complexity Penalty functions”.

Establishing dynamic characteristics of structures is a challenging area of research. The dynamic characteristics of structures, such as natural frequencies, modeshapes, response levels and damping characteristics play an important role in identifying the condition of the structures. The assumed modes method is a particular analytical method used to estimate the dynamic characteristics of a structure. However, the eigenfunctions used in the assumed mode method often led to ill-conditioning due to the presence of hyperbolic functions. Furthermore, a change in the boundary conditions of the system usually necessitates a change in the choice of assumed mode. In this thesis, a set of Alternative Admissible Functions (AAF), along with penalty functions, are used to obtain closed form solutions for an Euler-Bernoulli beam with various boundary conditions. A key advantage of the proposed approach is that the choice of AAF does not depend on the boundary conditions since the boundary conditions are modelled via penalty functions. The mathematical formulation is validated with different boundary conditions, Clamped-Free (CF), Simply-Supported (SS), and Clamped-Clamped (CC). A specific relation between the penalty function and the system parameters are established for CF, SS and CC boundary conditions to obtain appropriate values of penalties. Validation of results with the reported literature indicates excellent agreement when compared with closed-form Euler-Bernoulli beam values. The AAF approach with penalties is extended to a beam with a shallow crack to estimate the dynamic characteristics. The crack is modelled as a penalty function via a massless rotational spring. This model has the advantage of simplifying parametric studies, because of its discrete nature, allowing easy modification in the crack position and depth of the crack. Therefore, once the model is established, various practical applications may be performed without reformulation of the problem. Validation of results with the reported literature on beams with shallow cracks indicates the suitability of the proposed approach.

A partir d’un ensemble de données interprétées et issues d’une analyse préalable par un opérateur expert (horizons, failles), l’objectif de la thèse est de proposer une segmentation d’une image sismique sous-jacente en parfaite cohérence avec les lois de la géologie. L’originalité de la démarche consistera à développer des techniques de segmentation d’images sismiques, entre autres basées sur des approches de type contours actifs, contraintes par des données interprétées en supplément de propriétés intrinsèques calculées par des procédés automatiques à partir de la donnée traitée sans nécessiter une quelconque supervision contrairement aux travaux existants. Un deuxième axe consistera à ordonnancer automatiquement les horizons (surfaces) interprétés et analyser finement chaque intervalle (le lieu existant entre deux horizons), en prenant en compte son contenu (amplitude, orientation, etc.). Tout cela aboutissant à la reconstruction du pseudo-temps géologique. / The objective of the thesis is to propose a segmentation of an underlying seismic image in perfect coherence with the results of a preliminary analysis by an expert (horizons, faults). laws of geology. The originality of the approach will be to develop techniques for segmenting seismic images, among others based on active contour type approaches, constrained by data interpreted in addition to intrinsic properties calculated by automatic processes from the data processed without requiring any supervision in contrast to existing work. A second axis will be to automatically schedule the horizons (surfaces) interpreted and to analyze each interval (the place between two horizons) finely, taking into account its content (amplitude, orientation, etc.). All this resulted in the reconstruction of the geological pseudo-time.

Submitted by DAISY PAES SILVA null (daisypaess@gmail.com) on 2016-05-18T15:43:23Z No. of bitstreams: 1 Dissertação.pdf: 3068870 bytes, checksum: d65c9a34405a8cb377b1440005b0fb11 (MD5) / Approved for entry into archive by Ana Paula Grisoto (grisotoana@reitoria.unesp.br) on 2016-05-20T17:31:48Z (GMT) No. of bitstreams: 1 silva_dp_me_bauru.pdf: 3068870 bytes, checksum: d65c9a34405a8cb377b1440005b0fb11 (MD5) / Made available in DSpace on 2016-05-20T17:31:48Z (GMT). No. of bitstreams: 1 silva_dp_me_bauru.pdf: 3068870 bytes, checksum: d65c9a34405a8cb377b1440005b0fb11 (MD5) Previous issue date: 2016-03-29 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / O problema de Fluxo de Potência Ótimo (FPO) é considerado um importante problema da Engenharia Elétrica desde a década de 1960. A partir de então, muitos trabalhos foram publicados com diferentes formulações e abordagens para a resolução deste problema. Muitas destas abordagens desconsiderava a natureza discreta das variáveis de controle e consideram todas as variáveis do problema como contínuas. Estas formulações são aproximações do problema de FPO, pois, algumas variáveis podem somente ser ajustadas por passos discretos, conforme a realidade do sistema. No problema de Fluxo de Potência Ótimo Reativo (FPOR), caso particular do problema de FPO, as variáveis relacionadas à potência ativa são fixadas e a otimização somente considera as variáveis relacionadas à potência reativa. O problema de FPOR pode ser modelado matematicamente como um problema de programação não-linear com variáveis discretas e contínuas. Neste trabalho, propõem-se das abordagens para resolução do problema FPOR que consideram a natureza discreta das variáveis do problema. Nas abordagens propostas são utilizadas funções penalidade associadas a um método de pontos interiores, combinando as vantagens de ambos para a resolução do problema de FPOR. Desenvolvem-se funções penalidade polinomiais para tratar as variáveis de controle discretas do problema, taps dos transformadores e bancos de capacitores e de reatores shunt, obtendo-se uma sequência de problemas contínuos, diferenciáveis e penalizados, que são resolvidos pelo método de pontos interiores implementado no solver gratuito IPOPT. As soluções de tais problemas convergem para a solução do problema original. Os testes numéricos foram realizados com os sistemas elétricos IEEE 14, 30, 118 e 300 barras para verificar a eficiência das abordagens propostas. / The Optimal Power Flow Problem (OPF) is considered an important problem of the electrical engineering since the 1960s. From that moment, many papers were published with different formulations and approaches for solving this problem. Many of these approaches disregard the discrete nature of the control variables and consider all the variables of the problem as continuous. These formulations are approximations of the OPF problem, because some variables can be adjusted only by discrete steps, according to the system reality. In the Reactive Optimal Power Flow problem (ROPF), particular case of the OPF problem, the variables related to the active power are fixed and the optimization only considers the variables related to the reactive power. The ROPF problem can be mathematically modeled as a nonlinear programming problem with discrete and continuous variables. In this work, two approaches are presented for solving the ROPF problem considering the discrete nature of its variables. In the presented approaches, penalty functions are used associated with an interior-point method, combining the advantages of both for solving the ROPF problem. Polynomial penalty functions are used to treat the discrete control variables of the problem, transformers taps and shunt susceptances, obtaining a sequence of continuous, differentiable and penalized problems, which are solved by the interior-point method implemented in the IPOPT free solver. The solution of such problems converge to the solution of the original problem. The numerical tests were performed in the electrical systems IEEE 14, IEEE 30 and IEEE 118 buses to show the efficiency of the proposed methods. / CNPq: 130486/2014-0

Este trabalho aborda o problema de controle ótimo por modos deslizantes via função penalidade para sistemas de tempo discreto. Para resolver este problema será desenvolvido uma estrutura matricial alternativa baseada no problema de mínimos quadrados ponderados e funções penalidade. A partir desta nova formulação é possível obter a lei de controle ótimo por modos deslizantes, as equações de Riccati e a matriz do ganho de realimentação através desta estrutura matricial alternativa. A motivação para propormos essa nova abordagem é mostrar que é possível obter uma solução alternativa para o problema clássico de controle ótimo por modos deslizantes. / This work introduces a penalty function approach to deal with the optimal sliding mode control problem for discrete-time systems. To solve this problem an alternative array structure based on the problem of weighted least squares penalty function will be developed. Using this alternative matrix structure, the optimal sliding mode control law of, the matrix Riccati equations and feedback gain were obtained. The motivation of this new approach is to show that it is possible to obtain an alternative solution to the classic problem of optimal sliding mode control.

Este trabalho aborda o problema de controle ótimo por modos deslizantes via função penalidade para sistemas de tempo discreto. Para resolver este problema será desenvolvido uma estrutura matricial alternativa baseada no problema de mínimos quadrados ponderados e funções penalidade. A partir desta nova formulação é possível obter a lei de controle ótimo por modos deslizantes, as equações de Riccati e a matriz do ganho de realimentação através desta estrutura matricial alternativa. A motivação para propormos essa nova abordagem é mostrar que é possível obter uma solução alternativa para o problema clássico de controle ótimo por modos deslizantes. / This work introduces a penalty function approach to deal with the optimal sliding mode control problem for discrete-time systems. To solve this problem an alternative array structure based on the problem of weighted least squares penalty function will be developed. Using this alternative matrix structure, the optimal sliding mode control law of, the matrix Riccati equations and feedback gain were obtained. The motivation of this new approach is to show that it is possible to obtain an alternative solution to the classic problem of optimal sliding mode control.