by Wong Yin Bun, Terence. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1998. / Includes bibliographical references (leaves 99-104). / Abstract also in Chinese. / List of Figures --- p.ix / List of Tables --- p.xi / Chapter I --- Preliminary --- p.1 / Chapter 1 --- Introduction --- p.2 / Chapter 1.1 --- Thesis themes --- p.4 / Chapter 1.1.1 --- Dynamical view of landscape --- p.4 / Chapter 1.1.2 --- Bottom-up self-feedback algorithm with memory --- p.4 / Chapter 1.1.3 --- Cooperation and competition --- p.5 / Chapter 1.1.4 --- Contributions to genetic algorithms --- p.5 / Chapter 1.2 --- Thesis outline --- p.5 / Chapter 1.3 --- Contribution at a glance --- p.6 / Chapter 1.3.1 --- Problem --- p.6 / Chapter 1.3.2 --- Approach --- p.7 / Chapter 1.3.3 --- Contributions --- p.7 / Chapter 2 --- Background --- p.8 / Chapter 2.1 --- Iterative stochastic searching algorithms --- p.8 / Chapter 2.1.1 --- The algorithm --- p.8 / Chapter 2.1.2 --- Stochasticity --- p.10 / Chapter 2.2 --- Fitness landscapes and its relation to neighborhood --- p.12 / Chapter 2.2.1 --- Direct searching --- p.12 / Chapter 2.2.2 --- Exploration and exploitation --- p.12 / Chapter 2.2.3 --- Fitness landscapes --- p.13 / Chapter 2.2.4 --- Neighborhood --- p.16 / Chapter 2.3 --- Species formation methods --- p.17 / Chapter 2.3.1 --- Crowding methods --- p.17 / Chapter 2.3.2 --- Deterministic crowding --- p.18 / Chapter 2.3.3 --- Sharing method --- p.18 / Chapter 2.3.4 --- Dynamic niching --- p.19 / Chapter 2.4 --- Summary --- p.21 / Chapter II --- Probabilistic Binary Hierarchical Search --- p.22 / Chapter 3 --- The basic algorithm --- p.23 / Chapter 3.1 --- Introduction --- p.23 / Chapter 3.2 --- Search space reduction with binary hierarchy --- p.25 / Chapter 3.3 --- Search space modeling --- p.26 / Chapter 3.4 --- The information processing cycle --- p.29 / Chapter 3.4.1 --- Local searching agents --- p.29 / Chapter 3.4.2 --- Global environment --- p.30 / Chapter 3.4.3 --- Cooperative refinement and feedback --- p.33 / Chapter 3.5 --- Enhancement features --- p.34 / Chapter 3.5.1 --- Fitness scaling --- p.34 / Chapter 3.5.2 --- Elitism --- p.35 / Chapter 3.6 --- Illustration of the algorithm behavior --- p.36 / Chapter 3.6.1 --- Test problem --- p.36 / Chapter 3.6.2 --- Performance study --- p.38 / Chapter 3.6.3 --- Benchmark tests --- p.45 / Chapter 3.7 --- Discussion and analysis --- p.45 / Chapter 3.7.1 --- Hierarchy of partitions --- p.45 / Chapter 3.7.2 --- Availability of global information --- p.47 / Chapter 3.7.3 --- Adaptation --- p.47 / Chapter 3.8 --- Summary --- p.48 / Chapter III --- Cooperation and Competition --- p.50 / Chapter 4 --- High-dimensionality --- p.51 / Chapter 4.1 --- Introduction --- p.51 / Chapter 4.1.1 --- The challenge of high-dimensionality --- p.51 / Chapter 4.1.2 --- Cooperation - A solution to high-dimensionality --- p.52 / Chapter 4.2 --- Probabilistic Cooperative Binary Hierarchical Search --- p.52 / Chapter 4.2.1 --- Decoupling --- p.52 / Chapter 4.2.2 --- Cooperative fitness --- p.53 / Chapter 4.2.3 --- The cooperative model --- p.54 / Chapter 4.3 --- Empirical performance study --- p.56 / Chapter 4.3.1 --- pBHS versus pcBHS --- p.56 / Chapter 4.3.2 --- Scaling behavior of pcBHS --- p.60 / Chapter 4.3.3 --- Benchmark test --- p.62 / Chapter 4.4 --- Summary --- p.63 / Chapter 5 --- Deception --- p.65 / Chapter 5.1 --- Introduction --- p.65 / Chapter 5.1.1 --- The challenge of deceptiveness --- p.65 / Chapter 5.1.2 --- Competition: A solution to deception --- p.67 / Chapter 5.2 --- Probabilistic cooperative-competitive binary hierarchical search --- p.67 / Chapter 5.2.1 --- Overview --- p.68 / Chapter 5.2.2 --- The cooperative-competitive model --- p.68 / Chapter 5.3 --- Empirical performance study --- p.70 / Chapter 5.3.1 --- Goldberg's deceptive function --- p.70 / Chapter 5.3.2 --- "Shekel family - S5, S7, and S10" --- p.73 / Chapter 5.4 --- Summary --- p.74 / Chapter IV --- Finale --- p.78 / Chapter 6 --- A new genetic operator --- p.79 / Chapter 6.1 --- Introduction --- p.79 / Chapter 6.2 --- Variants of the integration --- p.80 / Chapter 6.2.1 --- Fixed-fraction-of-all --- p.83 / Chapter 6.2.2 --- Fixed-fraction-of-best --- p.83 / Chapter 6.2.3 --- Best-from-both --- p.84 / Chapter 6.3 --- Empricial performance study --- p.84 / Chapter 6.4 --- Summary --- p.88 / Chapter 7 --- Conclusion and Future work --- p.89 / Chapter A --- The pBHS Algorithm --- p.91 / Chapter A.1 --- Overview --- p.91 / Chapter A.2 --- Details --- p.91 / Chapter B --- Test problems --- p.96 / Bibliography --- p.99

Thesis (M. Phil.)--University of Hong Kong, 2007. / Title proper from title frame. Also available in printed format.

There are three families of exact methods used for probabilistic inference in belief nets. It is necessary to compare them and analyze the advantages and the disadvantages of each algorithm, and know the time cost of making inferences in a given belief network. This paper discusses the factors that influence the computation time of each algorithm, presents the predictive model of the time complexity for each algorithm and shows the statistical results of testing the algorithms with randomly generated belief networks. / Graduation date: 1991

Thesis (Ph. D.)--University of Texas at Austin, 2000. / Vita. Includes bibliographical references (leaves 318-335). Available also in a digital version from Dissertation Abstracts.

Thesis (doctoral)--Vereinigten Friedrichs-Universität Halle-Wittenberg, 1919. / Vita. Includes bibliographical references.

Academisch proefschrift--Amsterdam. / Summary in English. "Stellingen": 4 p. inserted. Bibliography: p. 90-91.

Inaug.-diss.--Berlin. / Lebanslauf. "Sendersbdruck aus den 'Schriften des Mathematischen institute und des institute für angewandte mathematik der Universität B̈erlin'/band 4." Includes bibliographical references.

In the field of underground radioactive waste disposal, complex computer models are used to describe the flow of groundwater through rocks. An important property in this context is transmissivity, the ability of the groundwater to pass through rocks, and the transmissivity field can be represented by a stochastic model. The stochastic model is included in complex computer models which determine the travel time for radionuclides released at one point to reach another. As well as the uncertainty due to the stochastic model, there may also be uncertainties in the inputs of these models. In order to quantify the uncertainties, Monte Carlo analyses are often used. However, for computationally expensive models, it is not always possible to obtain a large enough sample to provide accurate enough uncertainty analyses. In this thesis, we present the use of Bayesian emulation methodology as an alternative to Monte Carlo in the analysis of stochastic models. The idea behind Bayesian emulation methodology is that information can be obtained from a small number of runs of the model using a small sample from the input distribution. This information can then be used to make inferences about the output of the model given any other input. The current Bayesian emulation methodology is extended to emulate two statistics of a stochastic computer model; the mean and the distribution function of the output. The mean is a simple output statistic to emulate and provides some information about how the output changes due to changes in each input. The distribution function is more complex to emulate, however it is an important statistic since it contains information about the entire distribution of the outputs. Distribution functions of radionuclide travel times have been used as part of risk analyses for underground radioactive waste disposal. The extended methodology is presented using a case study.

The effectiveness of keyword-based search engines depends largely on the ability of a user to formulate proper queries that are both expressive and selective. However, web search queries issued by casual users are often short and with limited expressiveness. Query recommendation is a popular technique employed by search engines to help users refine their queries. Traditional similarity-based methods, however, often result in redundant and monotonic recommendations. We identify five basic requirements of a query recommendation system, namely relevancy, redundancy-free, diversity, ranking and efficiency. In particular, we focus on the requirements of redundancy-free and diversified recommendations. We propose the DQR framework, which mines a search log to achieve two goals: (1) It clusters search log queries to extract query concepts, based on which recommended queries are selected. Through query construction from the query concepts, we are able to avoid recommendation redundancy. (2) It employs a probabilistic model and a greedy heuristic algorithm to achieve recommendation diversification. Through a comprehensive user study we compare DQR against five other recommendation methods on real search log datasets. Our experiment shows that DQR outperforms the other methods in terms of relevancy, diversity, and ranking performance of the recommendations. At the same time, DQR also achieves high efficiency performance. / published_or_final_version / Computer Science / Master / Master of Philosophy

In "A Critique of Two Objective Probability Theories," I examine two extensionalist approaches to the analysis of objective probability, arguing ultimately that neither can succeed as analyses of objective probability. Beginning with extensional frequency analyses, I first examine the limiting frequency interpretation of Reichenbach and Salmon, arguing that it is unacceptable as it (1) fails to handle the single case--providing no basis for assigning a value other than 0 or 1; and (2) fails to provide a unique value for the probability as a limit of an infinite sequence--the problem of randomness. I further argue that references to "natural sequences" as a means of avoiding these problems must fail due to an interesting difficulty derived from special relativity. Turning next to Kyburg's finite frequency interpretation, I claim that while it incorporates certain gains within the extensional approach, it still succumbs to variations of the same problems inherent in the Reichenbach/Salmon interpretation. Kyburg's proposal, furthermore, is too narrow, not sufficiently encompassing the concept of objective probability desired. I conclude with an argument to the effect that no extensional frequency interpretation is able to provide an acceptable analysis of scientific conceptions of chance. I next consider a "propensity" interpretation provided by Mellor, which purports to provide an extensionalist analysis of objective chance on the basis of partial beliefs--i.e., a personalist framework. I argue that this approach fails because the dispositional (propensity) basis is an ad hoc addendum to what turns out to be merely a personalist theory. I then consider various alterations of Mellor's approach, with the conclusion that no such personalist-based approach is viable as an analysis of objective probability. I also examine Mellor's notion of dispositions, arguing that it is too deterministic, and that it must be replaced by a statistical notion better adapted to probability. Finally, these several considerations are taken both as a motivation for intensional frequency and propensity approaches, and as identifying certain pitfalls that any approach must guard against. In view of these findings, a rough outline of what would constitute an acceptable intensional frequency or propensity interpretation is indicated.