Thesis advisor: Zhushan Mandy Li / Differential item functioning (DIF) occurs when examinees of equal ability from different groups have different probabilities of correctly responding to certain items. DIF analysis aims to identify potentially biased items to ensure the fairness and equity of instruments, and has become a routine procedure in developing and improving assessments. This study proposed a DIF detection method using regularization techniques, which allows for simultaneous investigation of all items on a test for both uniform and nonuniform DIF. In order to evaluate the performance of the proposed DIF detection models and understand the factors that influence the performance, comprehensive simulation studies and empirical data analyses were conducted. Under various conditions including test length, sample size, sample size ratio, percentage of DIF items, DIF type, and DIF magnitude, the operating characteristics of three kinds of regularized logistic regression models: lasso, elastic net, and adaptive lasso, each characterized by their penalty functions, were examined and compared. Selection of optimal tuning parameter was investigated using two well-known information criteria AIC and BIC, and cross-validation. The results revealed that BIC outperformed other model selection criteria, which not only flagged high-impact DIF items precisely, but also prevented over-identification of DIF items with few false alarms. Among the regularization models, the adaptive lasso model achieved superior performance than the other two models in most conditions. The performance of the regularized DIF detection model using adaptive lasso was then compared to two commonly used DIF detection approaches including the logistic regression method and the likelihood ratio test. The proposed model was applied to analyzing empirical datasets to demonstrate the applicability of the method in real settings. / Thesis (PhD) — Boston College, 2019. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement and Evaluation.

La prédiction de connexions entre objets, basée soit sur une observation bruitée, soit sur une suite d'observations est un problème d'intérêt pour un nombre d'applications allant de la conception de système de recommandation en commerce électronique et réseaux sociaux jusqu'à l'inférence de réseaux en biologie moléculaire. Ce travail présente des formulations du problème de prédiction de lien, dans les cadres statique et temporel, comme un problème régularisé. Dans le scénario statique c'est la combinaison de deux normes bien connues, la norme L1 et la trace-norme qui permet de prédire les liens, alors que dans le cas dynamique, l'utilisation d'un modèle autoregressif sur des descripteurs linéaires permet d'améliorer la qualité de la prédiction. Nous étudierons la nature des solutions des problèmes d'optimisation à la fois en termes statistique et algorithmique. Des résultats empiriques encourageant mettent en évidence l'apport de la méthodologie adoptée. / Predicting connections among objects, based either on a noisy observation or on a sequence of observations, is a problem of interest for numerous applications such as recommender systems for e-commerce and social networks, and also in system biology, for inferring interaction patterns among proteins. This work presents formulations of the graph prediction problem, in both dynamic and static scenarios, as regularization problems. In the static scenario we encode the mixture of two different kinds of structural assumptions in a convex penalty involving the L1 and the trace norm. In the dynamic setting we assume that certain graph features, such as the node degree, follow a vector autoregressive model and we propose to use this information to improve the accuracy of prediction. The solutions of the optimization problems are studied both from an algorithmic and statistical point of view. Empirical evidences on synthetic and real data are presented showing the benefit of using the suggested methods.

Este trabalho estuda a técnica de reconstrução de imagens conhecido como tomografia por impedância elétrica em um domínio bidimensional. Esta técnica consiste na alocação de eletrodos na fronteira do volume e uma fonte injeta padrões de corrente através dos eletrodos e medem-se as voltagens resultantes na fronteira. Com estes dados estima-se a condutividade (ou resistividade) do interior do domínio criando-se uma imagem do mesmo. A tomografia por impedância elétrica é um problema inverso e mal posto no sentido de Hadamard. Estudam-se diversos métodos de solução para resolver o problema direto usando métodos numéricos como diferenças finitas e volumes finitos. Proporemos os métodos numéricos a serem aplicados na solução do problema direto os quais serão testados com problemas onde a solução analítica é conhecida. Posteriormente aplicaremos os métodos propostos ao problema especifico. Uma questão importante na reconstrução de imagens é propor a maneira como aproximar o Jacobiano (ou matriz de sensibilidade) do problema, assim desenvolvemos uma técnica para a aproximação do mesmo usando os dados fornecidos pelo problema direto. / In this work is studied the technique of reconstruction of images known as electrical impedance tomography for a two-dimensional domain. This technique consists in the allocation of electrodes on the border of the volume and a source injects patterns of current through the electrodes and then measuring voltages through the other electrodes. With these data it is estimated the conductivity (or resistivity) on the interior of the domain and an image is create of it. The electrical impedance tomography is an inverse and ill conditioned problem in the Hadamard sense. In this work, is studying some numerical methods to solve the direct problem and are applied numerical methods such as the finite difference method and the finite volume method. It is proposed some numerical methods to solve the direct problem which will be tested with analytical problems where the solution is known. Later, apply the methods proposed to the specific issue. An important issue in the reconstruction problems is about the Jacobian (or sensitivity matrix) aproximation, thus proposing a technique for the calculation of even using the data provided by the direct problem. Keywords:

Predicting connections among objects, based either on a noisy observation or on a sequence of observations, is a problem of interest for numerous applications such as recommender systems for e-commerce and social networks, and also in system biology, for inferring interaction patterns among proteins. This work presents formulations of the graph prediction problem, in both dynamic and static scenarios, as regularization problems. In the static scenario we encode the mixture of two different kinds of structural assumptions in a convex penalty involving the L1 and the trace norm. In the dynamic setting we assume that certain graph features, such as the node degree, follow a vector autoregressive model and we propose to use this information to improve the accuracy of prediction. The solutions of the optimization problems are studied both from an algorithmic and statistical point of view. Empirical evidences on synthetic and real data are presented showing the benefit of using the suggested methods.

This work is devoted to the convergence analysis of a modified Runge-Kutta-type iterative regularization method for solving nonlinear ill-posed problems under a priori and a posteriori stopping rules. The convergence rate results of the proposed method can be obtained under Hölder-type source-wise condition if the Fréchet derivative is properly scaled and locally Lipschitz continuous. Numerical results are achieved by using the Levenberg-Marquardt and Radau methods.

Neste trabalho apresenta-se uma discussão geral sobre problemas inversos, problemas mal-postos c técnicas de regularização, visando sua aplicabilidade em problemas térmicos. Métodos numéricos especiais são discutidos para a solução de problemas que apresentam instabilidade em relação aos dados. Tais métodos baseiam-se na utilização de restrições ou informações adicionais sobre a solução procurada. O problema de determinação da condição inicial da equação do calor é resolvido numericamente através destas técnicas, particularmente a regularização de Tikhonov e o príncipio da máxima entropia conectados ao príncipio da discrepância de Morozov são utilizados. / In this work we present a general discussion on invcrse problems, ill-posed problems and regularization techniqucs, applying these techniques to thermal problcms. Special numerical methods are discusscd in order to solve problerns for which the solution is unstable under data perturbations. Such methods are based on the utilization of restrictions or additional information on thc solution. The problern of determining the initial condition of thc heat equation is numerically solved beyond thesc techniques, particularly thc T ikhonov regularization and thc maximum entropy principie connected to thc Morozov's discrepancy principie are used.

Neste trabalho apresenta-se uma discussão geral sobre problemas inversos, problemas mal-postos c técnicas de regularização, visando sua aplicabilidade em problemas térmicos. Métodos numéricos especiais são discutidos para a solução de problemas que apresentam instabilidade em relação aos dados. Tais métodos baseiam-se na utilização de restrições ou informações adicionais sobre a solução procurada. O problema de determinação da condição inicial da equação do calor é resolvido numericamente através destas técnicas, particularmente a regularização de Tikhonov e o príncipio da máxima entropia conectados ao príncipio da discrepância de Morozov são utilizados. / In this work we present a general discussion on invcrse problems, ill-posed problems and regularization techniqucs, applying these techniques to thermal problcms. Special numerical methods are discusscd in order to solve problerns for which the solution is unstable under data perturbations. Such methods are based on the utilization of restrictions or additional information on thc solution. The problern of determining the initial condition of thc heat equation is numerically solved beyond thesc techniques, particularly thc T ikhonov regularization and thc maximum entropy principie connected to thc Morozov's discrepancy principie are used.

Este trabalho estuda a técnica de reconstrução de imagens conhecido como tomografia por impedância elétrica em um domínio bidimensional. Esta técnica consiste na alocação de eletrodos na fronteira do volume e uma fonte injeta padrões de corrente através dos eletrodos e medem-se as voltagens resultantes na fronteira. Com estes dados estima-se a condutividade (ou resistividade) do interior do domínio criando-se uma imagem do mesmo. A tomografia por impedância elétrica é um problema inverso e mal posto no sentido de Hadamard. Estudam-se diversos métodos de solução para resolver o problema direto usando métodos numéricos como diferenças finitas e volumes finitos. Proporemos os métodos numéricos a serem aplicados na solução do problema direto os quais serão testados com problemas onde a solução analítica é conhecida. Posteriormente aplicaremos os métodos propostos ao problema especifico. Uma questão importante na reconstrução de imagens é propor a maneira como aproximar o Jacobiano (ou matriz de sensibilidade) do problema, assim desenvolvemos uma técnica para a aproximação do mesmo usando os dados fornecidos pelo problema direto. / In this work is studied the technique of reconstruction of images known as electrical impedance tomography for a two-dimensional domain. This technique consists in the allocation of electrodes on the border of the volume and a source injects patterns of current through the electrodes and then measuring voltages through the other electrodes. With these data it is estimated the conductivity (or resistivity) on the interior of the domain and an image is create of it. The electrical impedance tomography is an inverse and ill conditioned problem in the Hadamard sense. In this work, is studying some numerical methods to solve the direct problem and are applied numerical methods such as the finite difference method and the finite volume method. It is proposed some numerical methods to solve the direct problem which will be tested with analytical problems where the solution is known. Later, apply the methods proposed to the specific issue. An important issue in the reconstruction problems is about the Jacobian (or sensitivity matrix) aproximation, thus proposing a technique for the calculation of even using the data provided by the direct problem. Keywords:

Neste trabalho apresenta-se uma discussão geral sobre problemas inversos, problemas mal-postos c técnicas de regularização, visando sua aplicabilidade em problemas térmicos. Métodos numéricos especiais são discutidos para a solução de problemas que apresentam instabilidade em relação aos dados. Tais métodos baseiam-se na utilização de restrições ou informações adicionais sobre a solução procurada. O problema de determinação da condição inicial da equação do calor é resolvido numericamente através destas técnicas, particularmente a regularização de Tikhonov e o príncipio da máxima entropia conectados ao príncipio da discrepância de Morozov são utilizados. / In this work we present a general discussion on invcrse problems, ill-posed problems and regularization techniqucs, applying these techniques to thermal problcms. Special numerical methods are discusscd in order to solve problerns for which the solution is unstable under data perturbations. Such methods are based on the utilization of restrictions or additional information on thc solution. The problern of determining the initial condition of thc heat equation is numerically solved beyond thesc techniques, particularly thc T ikhonov regularization and thc maximum entropy principie connected to thc Morozov's discrepancy principie are used.

No description available.