abstract: Bivariate responses that comprise mixtures of binary and continuous variables are common in medical, engineering, and other scientific fields. There exist many works concerning the analysis of such mixed data. However, the research on optimal designs for this type of experiments is still scarce. The joint mixed responses model that is considered here involves a mixture of ordinary linear models for the continuous response and a generalized linear model for the binary response. Using the complete class approach, tighter upper bounds on the number of support points required for finding locally optimal designs are derived for the mixed responses models studied in this work. In the first part of this dissertation, a theoretical result was developed to facilitate the search of locally symmetric optimal designs for mixed responses models with one continuous covariate. Then, the study was extended to mixed responses models that include group effects. Two types of mixed responses models with group effects were investigated. The first type includes models having no common parameters across subject group, and the second type of models allows some common parameters (e.g., a common slope) across groups. In addition to complete class results, an efficient algorithm (PSO-FM) was proposed to search for the A- and D-optimal designs. Finally, the first-order mixed responses model is extended to a type of a quadratic mixed responses model with a quadratic polynomial predictor placed in its linear model. / Dissertation/Thesis / Doctoral Dissertation Statistics 2020

Sound power measurements of acoustic sources are typically performed in anechoic or reverberation chambers using acoustic pressure according to international standards. The anechoic chamber creates a free-field environment where the sound power is estimated from the squared pressure integrated over some enveloping surface. The reverberation chamber produces diffuse-field conditions, where sound power is proportional to the spatially averaged squared pressure. In semi-reverberant environments, the direct and reverberant energies each contribute to the total measured field. If the kinetic and potential components of acoustic energy density are weighted appropriately, the spatial variation of the field can be significantly reduced compared to squared pressure. This generalized energy density allows an adaptation of the sound power formulation by Hopkins and Stryker to be used to make an efficient and accurate in situ sound power estimate of a noise source in a non-ideal acoustical environment. Since generalized energy density optimizes the spatial uniformity of the field, fewer measurement positions are needed compared to traditional standards. However, this method breaks down for sources that are large and extended in nature and considerably underestimates the sound power. This thesis explores the practical limits of this method related to the sound power underestimation. It also seeks to understand the special considerations necessary to achieve accurate, survey-grade sound power data of large, extended noise sources through a laboratory study of custom extended and compact sources. A modified method to accurately and efficiently measure the sound power of large, extended sources is proposed with results.

In this thesis, we propose to analyze panel count data using a spline-based sieve generalized estimating equation method with a semiparametric proportional mean model E(N(t)|Z) = Λ0(t) eβT0Z. The natural log of the baseline mean function, logΛ0(t), is approximated by a monotone cubic B-spline function. The estimates of regression parameters and spline coefficients are the roots of the spline based sieve generalized estimating equations (sieve GEE). The proposed method avoids assumingany parametric structure of the baseline mean function and the underlying counting process. Selection of an appropriate covariance matrix that represents the true correlation between the cumulative counts improves estimating efficiency. In addition to the parameters existing in the proportional mean function, the estimation that accounts for the over-dispersion and autocorrelation involves an extra nuisance parameter σ2, which could be estimated using a method of moment proposed by Zeger (1988). The parameters in the mean function are then estimated by solving the pseudo generalized estimating equation with σ2 replaced by its estimate, σ2n. We show that the estimate of (β0,Λ0) based on this two-stage approach is still consistent and could converge at the optimal convergence rate in the nonparametric/semiparametric regression setting. The asymptotic normality of the estimate of β0 is also established. We further propose a spline-based projection variance estimating method and show its consistency. Simulation studies are conducted to investigate finite sample performance of the sieve semiparametric GEE estimates, as well as different variance estimating methods with different sample sizes. The covariance matrix that accounts for the overdispersion generally increases estimating efficiency when overdispersion is present in the data. Finally, the proposed method with different covariance matrices is applied to a real data from a bladder tumor clinical trial.

A family of concave penalties, including the smoothly clipped absolute deviation (SCAD) and minimax concave penalties (MCP), has been shown to have attractive properties in variable selection. The computation of concave penalized solutions, however, is a difficult task. We propose a majorization minimization by coordinate descent (MMCD) algorithm to compute the solutions of concave penalized generalized linear models (GLM). In contrast to the existing algorithms that uses local quadratic or local linear approximation of the penalty, the MMCD majorizes the negative log-likelihood by a quadratic loss, but does not use any approximation to the penalty. This strategy avoids the computation of scaling factors in iterative steps, hence improves the efficiency of coordinate descent. Under certain regularity conditions, we establish the theoretical convergence property of the MMCD algorithm. We implement this algorithm in a penalized logistic regression model using the SCAD and MCP penalties. Simulation studies and a data example demonstrate that the MMCD works sufficiently fast for the penalized logistic regression in high-dimensional settings where the number of covariates is much larger than the sample size. Grouping structure among predictors exists in many regression applications. We first propose an l2 grouped concave penalty to incorporate such group information in a regression model. The l2 grouped concave penalty performs group selection and includes group Lasso as a special case. An efficient algorithm is developed and its theoretical convergence property is established under certain regularity conditions. The group selection property of the l2 grouped concave penalty is desirable in some applications; while in other applications selection at both group and individual levels is needed. Hence, we propose an l1 grouped concave penalty for variable selection at both individual and group levels. An efficient algorithm is also developed for the l1 grouped concave penalty. Simulation studies are performed to evaluate the finite-sample performance of the two grouped concave selection methods. The new grouped penalties are also used in analyzing two motivation datasets. The results from both the simulation and real data analyses demonstrate certain benefits of using grouped penalties. Therefore, the proposed concave group penalties are valuable alternatives to the standard concave penalties.

Syftet med denna studie var att lyfta fram pappors beskrivningar av att vara förälder utifrån deras förväntningar av vad en pappa respektive en man är. Detta för att få en överblick kring vilka förutsättningar och hinder män kan stå inför vid uttag av sin rätt till föräldraledighet. För att besvara frågeställningarna har ett kvalitativt metodologiskt angreppssätt använts i form av ostrukturerade intervjuer med öppna frågor. Urvalet i studien var sju fäder med minst ett barn under tre år. Intervjuerna genomfördes genom personligt möte med fem pappor och via telefon med de återstående papporna. De teman som diskuterades under intervjuerna var föräldraskap, arbete och försörjning, papparollen, mansrollen, föräldrastöd och förväntningar på sin partner och vice versa. Insamlad empiri analyserades utifrån en innehållsanalys där kodning och tematisering användes. I studien användes Raewyn Connells teori om olika maskuliniteter och George Herbert Meads teori om symboliska interaktionism som teoretiska utgångspunkter. Dessa teorier användes för att förstå mäns villkor i hur de kan utöva deras rätt till föräldraledighet. Detta genom att lyfta fram vilka barriärer män kan stå inför i förhållande till könsnormer, samt hur de hanterar dessa genom att belysa deras interaktion med andra människor. Resultatet visar att en destruktiv hegemonisk maskulinitet kan utgöra svårigheter för mäns föräldraskap, medan en uppmuntrande hegemonisk maskulinitet tillåter män att våga trotsa ideal om att de inte är hemma med sina barn. Med våra informanters berättelser kunde författarna skilja på en lojalitet gentemot informanternas arbetsplats som kan förklaras i Meads teori om den generaliserade andre. Interaktionen mellan kollegor och andra människor kan utgöra en tvingande arbetskultur som hindrar män att nyttja sina rättigheter. Studien framhäver att även om staten har upprättat reformer som främjar jämlikhet, till exempel de öronmärkta föräldradagarna, kan manlighetsideal förstås hindra staten från att nå sitt mål. / The purpose of this study was to highlight fathers’ descriptions of their expectations of what a father and a man are. The writers of this study aimed to get an overview of different conditions and barriers men can face when exercising their right to parental leave. To answer these questions was a qualitative method in terms of semi-structured interviews with open-ended questions used. The selection in the study was seven fathers with at least one child under three years. The interviews were conducted by a personal meeting with five of the fathers, while two interviews took place over the phone with the two remaining fathers. During the interview’s themes such as parenthood, employment and livelihood, the role of a father, the role of a man, parental support and expectations on their partner and vice versa were introduced. The empirical data is analyzed through a content analysis method, where coding and thematization were used. The study used Raewyn Connell’s theory of different masculinities and George Herbert Mead’s symbolic interaction as theoretical starting points. These theories were used to understand which barriers men face from gender norms and how they handle them being shown through interaction with other people. These conditions can avert men from exercising their right to parental leave. The result shows that a destructive hegemonic masculinity can pose difficulties for men’s parenthood, while an encouraging hegemonic masculinity allows men to dare to defy ideals about them not being home with their children. With our informant’s narrative we were able to distinguish a loyalty towards their workplace which could be emblematized in Mead’s notion about the generalized other. The interaction with colleagues and other people can constitute a coercive work culture that prevents men's protection of their rights. The study highlights that even though the state has established reforms that promote equality, for example the gender-segregated parental days, masculinity ideals can prevent the state from reaching its goal.

Cette thèse a pour sujet l'étude de la réaction de la Diffusion Compton Profondément Virtuelle (DVCS) sur le proton e + p -> e' + p' + gamma (DVCS). Cette réaction est mesurée en analysant l'expérience e1-dvcs2 qui a eu lieu dans le Hall B du Jefferson Laboratory avec CLAS. La prise de données s'est déroulée du 22 octobre 2008 jusqu'au 23 janvier 2009. Cette analyse a pour but la détermination des sections efficaces non polarisées et les différences de sections efficaces polarisées du DVCS, dans divers bins en xB, Q^2, t et phi. Cette analyse compare les sections efficaces avec plusieurs mesures existantes du DVCS: une analyse parallèle de e1-dvcs2 et une analyse de e1-dvcs1. En factorisant l'élément de matrice qui correspond au DVCS en deux parties, calculables en QCD perturbative et non perturbative, on peut accéder aux Distributions de Partons Généralisées (GPDs) du proton, et ainsi obtenir une vue en 3D des distributions spatiales et en impulsion des quarks et des gluons dans le proton. / This thesis focuses on the Deeply Virtual Compton Scattering (DVCS) reaction e + p -> e' + p' + gamma (DVCS). The reaction is measured using the e1-dvcs2 experiment run at Jefferson Laboratory in Hall B using CLAS. The experiment took place from 22 October, 2008 to 23 January, 2009, and experiment run time of 90 days. This analysis focuses on the determination of the DVCS cross section in bins of xB, Q^2, t and phi, and makes a comparison with already existing and parallel analyses of DVCS. By factorizing the cross section of the reaction into perturbative and non-perturbative parts, we may relate the cross section of this reaction to Generalized Parton Distributions (GPDs) for the proton, and in doing so, provide better insight as to the distributions of quarks and gluons within it, including spacial distribution and contributions of angular momentum.

abstract: Many real-world planning problems can be modeled as Markov Decision Processes (MDPs) which provide a framework for handling uncertainty in outcomes of action executions. A solution to such a planning problem is a policy that handles possible contingencies that could arise during execution. MDP solvers typically construct policies for a problem instance without re-using information from previously solved instances. Research in generalized planning has demonstrated the utility of constructing algorithm-like plans that reuse such information. However, using such techniques in an MDP setting has not been adequately explored. This thesis presents a novel approach for learning generalized partial policies that can be used to solve problems with different object names and/or object quantities using very few example policies for learning. This approach uses abstraction for state representation, which allows the identification of patterns in solutions such as loops that are agnostic to problem-specific properties. This thesis also presents some theoretical results related to the uniqueness and succinctness of the policies computed using such a representation. The presented algorithm can be used as fast, yet greedy and incomplete method for policy computation while falling back to a complete policy search algorithm when needed. Extensive empirical evaluation on discrete MDP benchmarks shows that this approach generalizes effectively and is often able to solve problems much faster than existing state-of-art discrete MDP solvers. Finally, the practical applicability of this approach is demonstrated by incorporating it in an anytime stochastic task and motion planning framework to successfully construct free-standing tower structures using Keva planks. / Dissertation/Thesis / Masters Thesis Computer Engineering 2020

京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第21540号 / 理博第4447号 / 新制||理||1639(附属図書館) / 京都大学大学院理学研究科数学・数理解析専攻 / (主査)教授 玉川 安騎男, 教授 向井 茂, 教授 望月 新一 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM

Change-talk (CT), or self-arguments for change, has been associated with favorable patient outcomes, while counter change-talk (CCT), or self-arguments against change, has been associated with poorer outcomes. Most studies on change language have focused on the prediction of distal posttreatment outcomes, while the prediction of more proximal outcomes has remained largely untested. Addressing this gap, we examined early treatment CT and CCT as predictors of worry change trajectories, “hazard” of clinically significant response, and time to response (i.e., outcome efficiency) in CBT and CBT integrated with MI (MI-CBT) for generalized anxiety disorder (GAD). We also explored whether treatment type moderated these associations. Data derived from a randomized controlled trial comparing CBT (n = 43) and MI-CBT (n = 42) for GAD. Independent observers reliably coded CT/CCT during session 1. Patients rated their worry after every session. Multilevel modeling revealed that, across both treatments, more CT associated with lower midtreatment worry level (p = .03), whereas more CCT associated with a slower rate of worry reduction at midtreatment (p = .04). However, treatment moderated the associations between CT and both midtreatment worry level (p = .004) and rate of change (p = .03). In CBT, patients with higher vs. lower CT had less worry and a faster rate of worry reduction; in MI-CBT, CT was unrelated to midtreatment worry level and the rate of worry change. Treatment did not moderate the CCT-worry relations. Survival analyses revealed that, across both treatments, more CT associated with a greater hazard of response (p = .004) and approached a faster time to response (p = .05), and more CCT associated with a lower hazard of response (p = .002) and approached a slower time to response (p = .06). Patient motivational language predicts proximal outcomes, and may be useful in differential treatment selection.

Aim: A randomized clinical trial demonstrated that responsively adding motivational interviewing (MI) to cognitive-behavioral therapy (CBT) for generalized anxiety disorder (GAD) outperformed CBT alone on long-term worry reduction (Westra et al., 2016). Consistent with MI’s additive aim, this effect was mediated by less patient midtreatment resistance in the integrative treatment (Constantino et al., 2019). Insofar as GAD is marked by interpersonal styles of excessive nonassertiveness and over accommodation, I tested here whether MI-CBT also outperformed CBT, across acute treatment and long-term follow up, on reducing these characteristic interpersonal problems. Moreover, as patient resistance is an interpersonal event for which person-centered MI should, according to theory, be more helpful than directive CBT, I tested if resistance also mediated the expected effect of treatment on the long-term interpersonal outcomes. Method: Eighty-five patients with severe GAD were randomly assigned to 15 sessions of MI-CBT or CBT. Patients completed a measure of interpersonal problems repeatedly through treatment and 12 months of follow up. Independent observers rated patient resistance at a midtreatment session. Results: As expected, structural equation models revealed comparable reductions in nonassertiveness and over accommodation across acute MI-CBT and CBT. Also as predicted, MI-CBT vs. CBT promoted significantly greater reduction in over accommodation problems over long-term follow up; however, this differential effect was only marginally significant for nonassertiveness problems. Finally, as predicted, the treatment effect on the level of both interpersonal problems at 12-month follow up was mediated by less midtreatment resistance in MI-CBT vs. CBT. Discussion: Results support that the benefit of adding MI to CBT for GAD extends to long-term interpersonal change, and they implicate resistance management as a candidate mechanism of this additive effect.