Edward Y. Chow and Alan S. Willsky. / "February 14, 1983." / Bibliography: p. 37-38. / Office of Naval Research Contract No. N00014-77-C-0224 NASA Ames Research Grant No. NGL-22-009-124

Thesis (Ph.D.)--University of Nebraska-Lincoln, 2008. / Title from title screen (site viewed Feb. 17, 2009). PDF text: x, 145 p. : ill. (some col.) ; 2 Mb. UMI publication number: AAT 3331439. Includes bibliographical references. Also available in microfilm and microfiche formats.

In this thesis, we consider a Bayesian bi-level variable selection problem in high-dimensional regressions. In many practical situations, it is natural to assign group membership to each predictor. Examples include that genetic variants can be grouped at the gene level and a covariate from different tasks naturally forms a group. Thus, it is of interest to select important groups as well as important members from those groups. The existing methods based on Markov Chain Monte Carlo (MCMC) are often computationally intensive and not scalable to large data sets. To address this problem, we consider variational inference for bi-level variable selection (BIVAS). In contrast to the commonly used mean-field approximation, we propose a hierarchical factorization to approximate the posterior distribution, by utilizing the structure of bi-level variable selection. Moreover, we develop a computationally efficient and fully parallelizable algorithm based on this variational approximation. We further extend the developed method to model data sets from multi-task learning. The comprehensive numerical results from both simulation studies and real data analysis demonstrate the advantages of BIVAS for variable selection, parameter estimation and computational efficiency over existing methods. The BIVAS software with support of parallelization is implemented in R package `bivas' available at https://github.com/mxcai/bivas.

Cette thèse traite du problème de la façon de prendre des décisions impliquant à la fois la temporisation et l'information ambiguë. Cette thèse se compose de trois chapitres. Le chapitre 1 passe en revue une série d'études sur l'influence de l'ambiguïté et de la temporisation sur la prise de décision individuelle, et soulève deux questions de recherche de la thèse actuelle: 1) Est-ce que les préférences d'ambiguïté des décideurs sont différentes pour les perspectives résolues dans le présent et l'avenir? et 2) Est-ce que les préférences temporelles de décideurs diffèrent sous les récompenses ambiguës et non ambiguës? Les chapitres 2 et 3 sont deux essais indépendants qui traitent de ces deux questions, respectivement. Le premier essai examine les préférences d'ambiguïté sous la résolution actuelle et retardée à travers les probabilités basses et hautes. Les résultats des trois études montrent un effet d'interaction entre le temps de résolution et le niveau de probabilité. Sous résolution immédiate, nous constatons que les individus présentent l'aversion d'ambiguïté à des probabilités élevées et ambiguïté-recherche faible, ou l'indifférence à faibles probabilités, cohérentes avec la littérature antérieure. Toutefois, la résolution future régresse aversion et de comportement de recherché à la neutralité. S’appuyant sur la théorie du niveau de construal et la théorie de double-processus, nous attribuons cet effet d'interaction à la différence de styles de traitement pour les perspectives présentes et futures. Le deuxième essai démontre l'impact de récompenses futures ambigus sur les préférences intertemporelles. Six études montrent que, malgré le fait que les récompenses ambiguës et retardées sont généralement detestés séparément, ensemble, elles produisent un effet positif. C'est-à-dire que, les récompenses ambiguës futures sont plus susceptibles d'être préférés que les récompenses précises (avec les valeurs attendues égales) dans la prise de decision intertemporelle. Nous proposons l'hypothèse de l’eclipse (overshadowing) pour expliquer cet effet et excluons trois autres possibilités. Enfin, nous établissons des conditions aux limites en examinant systématiquement si l'effet persiste à différents niveaux d'ambiguïté et de points de temps. / This dissertation addresses the issue of how to make decisions involving both time delay and ambiguous information. This dissertation is arranged into three chapters. Chapter 1 reviews a set of studies on the influence of ambiguity and time delay on individual decision making and raises two relevant research questions: (1) Are decision makers' ambiguity preferences different for prospects resolved in the present and the future?; and (2) Do decision makers' time preferences differ under ambiguous and unambiguous payoffs? Chapter 2 and 3 are two independent essays, each addressing one of the above questions. The first essay examines ambiguity preferences under present and delayed resolutions across low and high probabilities. Results of three studies show an interaction effect between resolution time and probability level. Under the immediate resolution, we find that individuals exhibit ambiguity aversion at high probabilities and weak ambiguity seeking or indifference at low probabilities, consistent with prior literature. However, delayed resolution regresses aversion and seeking behaviors to neutrality. Drawing on the construal level theory and the dual-process theory, we attribute this interaction effect to the difference in processing styles for present and future prospects. The second essay demonstrates the impact of ambiguous future payoffs on intertemporal preferences. Six studies show that, despite the fact that ambiguous and delayed payoffs are generally disliked separately, together they produce a positive effect. That is, ambiguous future payoffs are more likely to be preferred than precise payoffs (with equal expected values) in intertemporal decision-making. We propose the overshadowing hypothesis to explain this effect and rule out three other possibilities. Finally, we establish boundary conditions by systematically examining whether the effect persists at various ambiguity levels and time points.

The elderly living in smart homes can have their daily movement recorded and analyzed. Given the fact that different elders can have their own living habits, a methodology that can automatically identify their daily activities and discover their daily routines will be useful for better elderly care and support. In this thesis research, we focus on developing data mining algorithms for automatic detection of behavioral patterns from the trajectory data of an individual for activity identification, daily routine discovery, and activity prediction. The key challenges for the human activity analysis include the need to consider longer-range dependency of the sensor triggering events for activity modeling and to capture the spatio-temporal variations of the behavioral patterns exhibited by human. We propose to represent the trajectory data using a behavior-aware flow graph which is a probabilistic finite state automaton with its nodes and edges attributed with some local behavior-aware features. Subflows can then be extracted from the flow graph using the kernel k-means as the underlying behavioral patterns for activity identification. Given the identified activities, we propose a novel nominal matrix factorization method under a Bayesian framework with Lasso to extract highly interpretable daily routines. To better take care of the variations of activity durations within each daily routine, we further extend the Bayesian framework with a Markov jump process as the prior to incorporate the shift-invariant property into the model. For empirical evaluation, the proposed methodologies have been compared with a number of existing activity identification and daily routine discovery methods based on both synthetic and publicly available real smart home data sets with promising results obtained. In the thesis, we also illustrate how the proposed unsupervised methodology could be used to support exploratory behavior analysis for elderly care.

abstract: This thesis seeks to defend transitivity as a rational constraint on preferences against two putative counterexamples to transitivity. This thesis is divided into three sections. In the first section, I consider two famous and popular arguments in defense of transitivity and argue they are insufficient to adequately defend transitivity. I then outline a desiderata for successful arguments in defense of transitivity and identify some basic assumptions I will be making throughout the thesis. In section two, I consider the first putative counterexample to transitivity: Quinn’s Puzzle of the Self-Torturer. I offer two plausible interpretations of Quinn’s puzzle and argue that both fail. One because it does not genuinely induce intransitive preferences, and the other because the situation it requires is logically impossible. I conclude this section by defending my arguments against known objections in the literature. Finally, in the third section, I consider a counterexample to transitivity from Larry Temkin that has received little attention in the literature. I argue that while the initial counterexample is unpersuasive it can be augmented and made into a more forceful argument. I then argue that this improved counterexample fails due to some erroneous assumptions prevalent in the literature on incomparability. I conclude the thesis with a brief summary and some closing remarks. / Dissertation/Thesis / Masters Thesis Philosophy 2015

Esta tese reformula o modelo de teoria de decisão de Savage relaxando a hipótese implícita de que uma conseqüência é uma descrição perfeita de uma determinada situação. Axiomas comportamentais sobre preferências definidas no espaço de atos são introduzidos e uma representação na forma de Utilidade Esperada é derivada. Em particular, como em Savage, há uma única probabilidade subjetiva sobre os estados da natureza. O ganho de flexibilidade da reformulação apresenta uma solução para o paradoxo de Ellsberg que não faz uso de múltiplas probabilidades subjetivas, e uma reinterpretação da aversão ao risco no modelo de Utilidade Esperada convencional. / This thesis reformulates the Savage\'s Decision Theory model relaxing the implicit hypothesis that a consequence must be a perfect description of a situation. We introduce Behavioral axioms on preferences defined over the set of acts and derive a new Expected Utility functional representation. Like on Savage\'s, there is an unique prior over states of world. The flexibility gain of this representation presents a solution for the Ellsberg paradox that does not rely on multiple priors, and allows for a new interpretation of risk aversion on the Expect Utility model.

University of Macau / Faculty of Science and Technology / Department of Civil and Environmental Engineering

The secondary structure of proteins is predicted using various binary classifiers. The data are adopted from the RS126 database. The original data consists of protein primary and secondary structure sequences. The original data is encoded using alphabetic letters. These data are encoded into unary vectors comprising ones and zeros only. Different binary classifiers, namely the naive Bayes, logistic regression and classification trees using hold-out and 5-fold cross validation are trained using the encoded data. For each of the classifiers three classification tasks are considered, namely helix against not helix (H/∼H), sheet against not sheet (S/∼S) and coil against not coil (C/∼C). The performance of these binary classifiers are compared using the overall accuracy in predicting the protein secondary structure for various window sizes. Our result indicate that hold-out cross validation achieved higher accuracy than 5-fold cross validation. The Naive Bayes classifier, using 5-fold cross validation achieved, the lowest accuracy for predicting helix against not helix. The classification tree classifiers, using 5-fold cross validation, achieved the lowest accuracies for both coil against not coil and sheet against not sheet classifications. The logistic regression classier accuracy is dependent on the window size; there is a positive relationship between the accuracy and window size. The logistic regression classier approach achieved the highest accuracy when compared to the classification tree and Naive Bayes classifiers for each classification task; predicting helix against not helix with accuracy 77.74 percent, for sheet against not sheet with accuracy 81.22 percent and for coil against not coil with accuracy 73.39 percent. It is noted that it is easier to compare classifiers if the classification process could be completely facilitated in R. Alternatively, it would be easier to assess these logistic regression classifiers if SPSS had a function to determine the accuracy of the logistic regression classifier.

Hierarchical task decompositions play an essential role in the design of complex simulation and decision systems, such as the ones that arise in video games. Game designers find it very natural to adopt a divide-and-conquer philosophy of specifying hierarchical policies, where decision modules can be constructed somewhat independently. The process of choosing the parameters of these modules manually is typically lengthy and tedious. The hierarchical reinforcement learning (HRL) field has produced elegant ways of decomposing policies and value functions using semi-Markov decision processes. However, there is still a lack of demonstrations in larger nonlinear systems with discrete and continuous variables. To narrow this gap between industrial practices and academic ideas, we address the problem of designing efficient algorithms to facilitate the deployment of HRL ideas in more realistic settings. In particular, we propose Bayesian active learning methods to learn the relevant aspects of either policies or value functions by focusing on the most relevant parts of the parameter and state spaces respectively. To demonstrate the scalability of our solution, we have applied it to The Open Racing Car Simulator (TORCS), a 3D game engine that implements complex vehicle dynamics. The environment is a large topological map roughly based on downtown Vancouver, British Columbia. Higher level abstract tasks are also learned in this process using a model-based extension of the MAXQ algorithm. Our solution demonstrates how HRL can be scaled to large applications with complex, discrete and continuous non-linear dynamics. / Science, Faculty of / Computer Science, Department of / Graduate