Dans cette thèse, nous nous intéressons à l'estimation de paramètres dans les chaînes de Markov cachées. Nous considérons tout d'abord le problème de l'estimation en ligne (sans sauvegarde des observations) au sens du maximum de vraisemblance. Nous proposons une nouvelle méthode basée sur l'algorithme Expectation Maximization appelée Block Online Expectation Maximization (BOEM). Cet algorithme est défini pour des chaînes de Markov cachées à espace d'état et espace d'observations généraux. Dans le cas d'espaces d'états généraux, l'algorithme BOEM requiert l'introduction de méthodes de Monte Carlo séquentielles pour approcher des espérances sous des lois de lissage. La convergence de l'algorithme nécessite alors un contrôle de la norme Lp de l'erreur d'approximation Monte Carlo explicite en le nombre d'observations et de particules. Une seconde partie de cette thèse se consacre à l'obtention de tels contrôles pour plusieurs méthodes de Monte Carlo séquentielles. Nous étudions enfin des applications de l'algorithme BOEM à des problèmes de cartographie et de localisation simultanées. La dernière partie de cette thèse est relative à l'estimation non paramétrique dans les chaînes de Markov cachées. Le problème considéré est abordé dans un cadre précis. Nous supposons que (Xk) est une marche aléatoire dont la loi des incréments est connue à un facteur d'échelle a près. Nous supposons que, pour tout k, Yk est une observation de f(Xk) dans un bruit additif gaussien, où f est une fonction que nous cherchons à estimer. Nous établissons l'identifiabilité du modèle statistique et nous proposons une estimation de f et de a à partir de la vraisemblance par paires des observations. / This document is dedicated to inference problems in hidden Markov models. The first part is devoted to an online maximum likelihood estimation procedure which does not store the observations. We propose a new Expectation Maximization based method called the Block Online Expectation Maximization (BOEM) algorithm. This algorithm solves the online estimation problem for general hidden Markov models. In complex situations, it requires the introduction of Sequential Monte Carlo methods to approximate several expectations under the fixed interval smoothing distributions. The convergence of the algorithm is shown under the assumption that the Lp mean error due to the Monte Carlo approximation can be controlled explicitly in the number of observations and in the number of particles. Therefore, a second part of the document establishes such controls for several Sequential Monte Carlo algorithms. This BOEM algorithm is then used to solve the simultaneous localization and mapping problem in different frameworks. Finally, the last part of this thesis is dedicated to nonparametric estimation in hidden Markov models. It is assumed that the Markov chain (Xk) is a random walk lying in a compact set with increment distribution known up to a scaling factor a. At each time step k, Yk is a noisy observations of f(Xk) where f is an unknown function. We establish the identifiability of the statistical model and we propose estimators of f and a based on the pairwise likelihood of the observations.

The research explores the escalation of the herder-farmer conflict in Nigeria to identify the significant patterns of escalation. Relying on a mixed-method analysis of secondary data and aligning with the analytical anchorage of dynamic systems theory, the research argues that the although Benue and Enugu observe the same herder-farmer the patterns of conflict escalation is neither similar, linear or recurrent. This research submits ethnoreligious antagonism, lawlessness, and exclusionary politics as reasons why the conflict escalated into widespread violence. Adding that each of these elements self- reinforces and influence each other to sustain a coordinated state of violence or maintain peace. It makes a case for pragmatic policies that captures the history and political, economic, and social interaction of states and local government.

No description available.

This study examined the existence of implicit racial bias among public school teachers within the gifted referral process. Public school teachers from urban, suburban, and rural school districts surrounding a large northeastern city were be provided vignettes of gifted students demonstrating “typical” and “hidden” giftedness. The names and races of students within the vignettes were randomized to represent either a White male student or a Black male student. Univariate and multivariate analyses were utilized to determine the existence of significant differences in perceptions of giftedness and need for referral among teachers. In contrast to the hypotheses of the study, vignettes describing Black “typically” gifted students were rated as significantly higher than White “typically” gifted students. Black students also did not experience a significant decrease in ratings of giftedness and need for referral when described as “hidden” gifted. Lastly, results demonstrated a significant interaction where White students experienced a significant increase in both ratings of giftedness and need for referral when described as showing signs of “hidden” giftedness compared to their White “typically” gifted counterparts. Further discussion of these results along with imitations and considerations, most importantly the presence of social desirability bias, can be found at the end of this work. / School Psychology

Throughout Christianity's tumultuous history in Japan, there have been several traditions which have stood independent from Western missionary churches. Two such traditions are the Kakure Kirishitan (“Hidden Christians”) and Uchimura Kanzo's Non-Church Movement. Both have interpreted Christianity in ways that make sense within their own historical and cultural contexts. Japan's Hidden Christians were forced by strict persecution to practice their faith in secret, where they developed ways to disguise their practices. Meanwhile, at the dawn of a new era of religious freedom in Japan, Uchimura Kanzo formed a new way to practice Christianity that both integrated Japanese traditions such as bushido while rejecting typical structures of church hierarchy and organization. Through this project I hope to give a voice and proper agency to the often overlooked indigenized ways of practicing Christianity. Japanese Christian communities have forged their own religious practices that force us to expand our understanding of what it means to be Christian and what Christianity can look like in the lives of everyday people. The focus shifts away from church authorities and dogmatic proclamations, thus empowering and recognizing the authority of lay practitioners to make their own meaning from the Christian tradition.

Critical aspects of spacecraft missions, such as component organization, control algorithms, and trajectories, can be optimized using a variety of algorithms or solvers. Each solver has intrinsic strengths and weaknesses when applied to a given optimization problem. One way to mitigate limitations is to combine different solvers in an island model that allows these algorithms to share solutions. The program Spacecraft Trajectory Optimization Suite (STOpS) is an island model suite of heterogeneous and homogeneous Evolutionary Algorithms (EA) that analyze interplanetary trajectories for multiple gravity assist (MGA) missions. One limitation of STOpS and other spacecraft trajectory optimization programs (GMAT and Pygmo/Pagmo) is that they require a defined encounter body sequence to produce a constant length set of design variables. Early phase trajectory design would benefit from the ability to consider problems with an undefined encounter sequence as it would provide a set of diverse trajectories -- some of which might not have been considered during mission planning. The Hybrid Optimal Control Problem (HOCP) and the concept of hidden genes are explored with the most common EA, the Genetic Algorithm (GA), to compare how the methods perform with a Variable Size Design Space (VSDS). Test problems are altered so that the input to the cost function (the object being optimized) contains a set of continuous variables whose length depends on a corresponding set of discrete variables (e.g. the number of planet encounters determines the number of transfer time variables). Initial testing with a scalable problem (Branin's function) indicates that even though the HOCP consistently converges on an optimal solution, the expensive run time (due to algorithm collaboration) would only escalate in an island model system. The hidden gene mechanism only changes how the GA decodes variables, thus it does not impact run time and operates effectively in the island model. A Hidden Gene Genetic Algorithm ( HGGA) is tested with a simplified Mariner 10 (EVM) problem to determine the best parameter settings to use in an island model with the GTOP Cassini 1 (EVVEJS) problem. For an island model with all GAs there is improved performance when the different base algorithm settings are used. Similar to previous work, the model benefits from migration of solutions and using multiple algorithms or islands. For spacecraft trajectory optimization programs that have an unconstrained fly-by sequence, the design variable limits have the largest impact on the results. When the number of potential fly-by sequences is too large it prevents the solver from converging on an optimal solution. This work demonstrates HGGA is effective in the STOpS environment as well as with GTOP problems. Thus the hidden gene mechanism can be extended to other EAs with members containing design variables that function similarly. It is shown that the tuning of the HGGA is dependent on the specific constraints of the spacecraft trajectory problem at hand, thus there is no need to further explore the general capabilities of the algorithm.

We address the problem of learning discrete hidden Markov models from very long sequences of observations. Incremental versions of the Baum-Welch algorithm that approximate the beta-values used in the backward procedure are commonly used for this problem since their memory complexity is independent of the sequence length. However, traditional approaches have two main disadvantages: the approximation of the beta-values deviates far from the real values, and the learning algorithm requires previous knowledge of the topology of the model. This dissertation describes a new incremental Baum-Welch algorithm with a novel backward procedure that improves the approximation of the â-values based on a one-step lookahead in the training sequence and investigates heuristics to prune unnecessary states from an initial complex model. Two new approaches for pruning, greedy and controlled, are introduced and a novel method for identification of ill-conditioned models is presented. Incremental learning of multiple independent observations is also investigated. We justify the new approaches analytically and report empirical results that show they converge faster than the traditional Baum-Welch algorithm using fewer computer resources. Furthermore, we demonstrate that the new learning algorithms converge faster than the previous incremental approaches and can be used to perform online learning of high-quality models useful for classification tasks. Finally, this dissertation explores the use of the new algorithms for anomaly detection in computer systems, that improve our previous research work on detectors based on hidden Markov models integrated into real-world monitoring systems of high-performance computers.

No description available.

No description available.

No description available.