Thesis (M.S.)--Miami University, Dept. of Physics, 2009. / Title from first page of PDF document. Includes bibliographical references (p. 57-61).

abstract: As pointed out in the keynote speech by H. V. Jagadish in SIGMOD'07, and also commonly agreed in the database community, the usability of structured data by casual users is as important as the data management systems' functionalities. A major hardness of using structured data is the problem of easily retrieving information from them given a user's information needs. Learning and using a structured query language (e.g., SQL and XQuery) is overwhelmingly burdensome for most users, as not only are these languages sophisticated, but the users need to know the data schema. Keyword search provides us with opportunities to conveniently access structured data and potentially significantly enhances the usability of structured data. However, processing keyword search on structured data is challenging due to various types of ambiguities such as structural ambiguity (keyword queries have no structure), keyword ambiguity (the keywords may not be accurate), user preference ambiguity (the user may have implicit preferences that are not indicated in the query), as well as the efficiency challenges due to large search space. This dissertation performs an expansive study on keyword search processing techniques as a gateway for users to access structured data and retrieve desired information. The key issues addressed include: (1) Resolving structural ambiguities in keyword queries by generating meaningful query results, which involves identifying relevant keyword matches, identifying return information, composing query results based on relevant matches and return information. (2) Resolving structural, keyword and user preference ambiguities through result analysis, including snippet generation, result differentiation, result clustering, result summarization/query expansion, etc. (3) Resolving the efficiency challenge in processing keyword search on structured data by utilizing and efficiently maintaining materialized views. These works deliver significant technical contributions towards building a full-fledged search engine for structured data. / Dissertation/Thesis / Ph.D. Computer Science 2011

abstract: Multi-task learning (MTL) aims to improve the generalization performance (of the resulting classifiers) by learning multiple related tasks simultaneously. Specifically, MTL exploits the intrinsic task relatedness, based on which the informative domain knowledge from each task can be shared across multiple tasks and thus facilitate the individual task learning. It is particularly desirable to share the domain knowledge (among the tasks) when there are a number of related tasks but only limited training data is available for each task. Modeling the relationship of multiple tasks is critical to the generalization performance of the MTL algorithms. In this dissertation, I propose a series of MTL approaches which assume that multiple tasks are intrinsically related via a shared low-dimensional feature space. The proposed MTL approaches are developed to deal with different scenarios and settings; they are respectively formulated as mathematical optimization problems of minimizing the empirical loss regularized by different structures. For all proposed MTL formulations, I develop the associated optimization algorithms to find their globally optimal solution efficiently. I also conduct theoretical analysis for certain MTL approaches by deriving the globally optimal solution recovery condition and the performance bound. To demonstrate the practical performance, I apply the proposed MTL approaches on different real-world applications: (1) Automated annotation of the Drosophila gene expression pattern images; (2) Categorization of the Yahoo web pages. Our experimental results demonstrate the efficiency and effectiveness of the proposed algorithms. / Dissertation/Thesis / Ph.D. Computer Science 2011

Natural Language Generation (NLG) is the task of generating natural language (e.g., English sentences) from machine readable input. In the past few years, deep neural networks have received great attention from the natural language processing community due to impressive performance across different tasks. This thesis addresses NLG problems with deep neural networks from two different modeling views. Under the first view, natural language sentences are modelled as sequences of words, which greatly simplifies their representation and allows us to apply classic sequence modelling neural networks (i.e., recurrent neural networks) to various NLG tasks. Under the second view, natural language sentences are modelled as dependency trees, which are more expressive and allow to capture linguistic generalisations leading to neural models which operate on tree structures. Specifically, this thesis develops several novel neural models for natural language generation. Contrary to many existing models which aim to generate a single sentence, we propose a novel hierarchical recurrent neural network architecture to represent and generate multiple sentences. Beyond the hierarchical recurrent structure, we also propose a means to model context dynamically during generation. We apply this model to the task of Chinese poetry generation and show that it outperforms competitive poetry generation systems. Neural based natural language generation models usually work well when there is a lot of training data. When the training data is not sufficient, prior knowledge for the task at hand becomes very important. To this end, we propose a deep reinforcement learning framework to inject prior knowledge into neural based NLG models and apply it to sentence simplification. Experimental results show promising performance using our reinforcement learning framework. Both poetry generation and sentence simplification are tackled with models following the sequence learning view, where sentences are treated as word sequences. In this thesis, we also explore how to generate natural language sentences as tree structures. We propose a neural model, which combines the advantages of syntactic structure and recurrent neural networks. More concretely, our model defines the probability of a sentence by estimating the generation probability of its dependency tree. At each time step, a node is generated based on the representation of the generated subtree. We show experimentally that this model achieves good performance in language modeling and can also generate dependency trees.

This thesis presents an empirical research that aims to explore insight generation in discrete-event simulation (DES) studies. It is often claimed that simulation is useful for generating insights. There is, however, almost no empirical evidence to support this claim. The factors of a simulation intervention that affect the occurrence of insight are not clear. A specific claim is that watching the animated display of a simulation model is more helpful in making better decisions than relying on the statistical outcomes generated from simulation runs; but again, there is very limited evidence to support this. To address this dearth of evidence, two studies are implemented: a quantitative and a qualitative study. In the former, a laboratory-based experimental study is used, where undergraduate students were placed in three separate groups and given a task to solve using a model with only animation, a model with only statistical results, or using no model at all. In the qualitative study, semi-structured interviews with simulation consultants were carried out, where participants were requested to account examples of projects in which clients change their problem understanding and generate more effective ideas. The two separated parts of the study found different types of evidence to support that simulation generates insight. The experimental study suggests that insights are generated more rapidly from statistical results than the use of animation. Research outcomes from the interviews include descriptions of: the phase of a simulation study where insight emerges; the role of different methods applied and means used in discovering and overcoming discontinuity in thinking (for instance, the role of consultant s influence in problem understanding); how some factors of a simulation intervention are associated with the processes of uncovering and overcoming discontinuity in thinking (for example, the role of clients team in the selection of methods used to communicate results); and the role of the model and consultant in generating new ideas. This thesis contributes to the limited existing literature by providing a more in depth understanding of insight in the context of simulation and empirical evidence on the insight-enabling benefits of simulation based on an operational definition. The findings of the study provide new insights into the factors of simulation that support fast and creative problem solving.

Récemment, de nombreuses techniques de microscopie de fluorescence de super-résolution ont été développées pour permettre d'observer de nombreuses structures biologiques au-delà de la limite de diffraction. La microscopie d'illumination structurée (SIM) est l'une de ces technologies. Le principe de la SIM est basé sur l'utilisation d'une grille de lumière harmonique qui permet de translater les hautes fréquences spatiales de l'échantillon vers la région d’observation du microscope. Les méthodes classiques de reconstruction SIM nécessitent une connaissance parfaite de l'illumination de l’échantillon. Cependant, l’implémentation d’un contrôle parfait de l’illumination harmonique sur le plan de l'échantillon n'est pas facile expérimentalement et il présente un grand défi. L’hypothèse de la connaissance parfaite de l’intensité de la lumière illuminant l’échantillon en SIM peut donc introduire des artefacts sur l’image reconstruite de l'échantillon, à cause des erreurs d’alignement de la grille qui peuvent se présenter lors de l’acquisition expérimentale. Afin de surmonter ce défi, nous avons développé dans cette thèse des stratégies de reconstruction «aveugle» qui sont indépendantes de d'illumination. À l'aide de ces stratégies de reconstruction dites «blind-SIM», nous avons étendu la SIM harmonique pour l’appliquer aux cas de «SIM-speckle» qui utilisent des illuminations aléatoires et inconnues qui contrairement à l’illumination harmonique, ne nécessitent pas de contrôle. / Recently, many superresolution fluorescence microscopy techniques have been developed which allow the observation of many biological structures beyond the diffraction limit. Structured illumination microscopy (SIM) is one of them. The principle of SIM is based on using a harmonic light grid which downmodulates the high spatial frequencies of the sample into the observable region of the microscope. The resolution enhancement is highly dependent on the reconstruction technique, which restores the high spatial frequencies of the sample to their original position. Common SIM reconstructions require the perfect knowledge of the illumination pattern. However, to perfectly control the harmonic illumination patterns on the sample plane is not easy in experimental implementations and this makes the experimental setup very technical. Reconstructing SIM images assuming the perfect knowledge of the illumination intensity patterns may, therefore, introduce artifacts on the estimated sample due to the misalignment of the grid that can occur during experimental acquisitions. To tackle this drawback of SIM, in this thesis, we have developed blind-SIM reconstruction strategies which are independent of the illumination patterns. Using the 3D blind-SIM reconstruction strategies we extended the harmonic SIM to speckle illumination microscopy which uses random unknown speckle patterns that need no control, unlike the harmonic grid patterns.

abstract: Sparsity has become an important modeling tool in areas such as genetics, signal and audio processing, medical image processing, etc. Via the penalization of l-1 norm based regularization, the structured sparse learning algorithms can produce highly accurate models while imposing various predefined structures on the data, such as feature groups or graphs. In this thesis, I first propose to solve a sparse learning model with a general group structure, where the predefined groups may overlap with each other. Then, I present three real world applications which can benefit from the group structured sparse learning technique. In the first application, I study the Alzheimer's Disease diagnosis problem using multi-modality neuroimaging data. In this dataset, not every subject has all data sources available, exhibiting an unique and challenging block-wise missing pattern. In the second application, I study the automatic annotation and retrieval of fruit-fly gene expression pattern images. Combined with the spatial information, sparse learning techniques can be used to construct effective representation of the expression images. In the third application, I present a new computational approach to annotate developmental stage for Drosophila embryos in the gene expression images. In addition, it provides a stage score that enables one to more finely annotate each embryo so that they are divided into early and late periods of development within standard stage demarcations. Stage scores help us to illuminate global gene activities and changes much better, and more refined stage annotations improve our ability to better interpret results when expression pattern matches are discovered between genes. / Dissertation/Thesis / Ph.D. Computer Science 2013

The healthcare today is experiening a greater burden since diseases suchas cancer are more common. The diagnostic parts of the healthcare, suchas radiology and pathology, are aected with increased workload. Duringthe past several decades, systems for structured reporting in radiology havebecome available in a try to facilitate their work ow. The introduction ofdigital pathology has enabled the possibility to introduce structured reportingin pathology as well. The question is whether it can facilitate their work ow.Today's aids for structured reporting in radiology are more or less perceivedas distracting, and the challenge in this thesis is to create an aid for structuredreporting that is not distracting the pathologist's diagnostic work ow.To achieve this, a prototype with a template for invasive breast cancer andprostate cancer was implemented in Sectra's viewer for pathology images. Thetemplate for invasive breast cancer was tested by two pathologists in a userstudy with the main objective to determine the dierences in the diagnosticwork ow using the prototype and using only paper and pen. The pathologistcould see a use of the prototype both for breast assessment and assessmentsin other areas of pathology. Both pathologists also think that the prototypewill save time in their overall work ow, help them organize the informationretrieved during the assessment, and create an overall better diagnostic work- ow.

Doctor of Philosophy / Department of Educational Leadership / Sarah Jane Fishback / This exploratory phenomenological research study describes the experiences of U.S. Army soldiers going through the mandated Structured Self Development (SSD) online courseware. Multiple findings are presented covering soldier participants’ experiences with the process, content, and culture/environment of SSD. Additionally, findings dealing with soldiers’ motivations and self-described impediments while going through SSD are presented. Four Army enlisted soldiers (two male, two female) were purposefully selected for this study, each one representing a different level of SSD (Levels 1-4). Participant soldiers for this study were selected from throughout the Kansas National Guard and each one possessed a different duty military occupational specialty within the Army. The findings of the research study indicate that there are multiple aspects of SSD that soldiers experienced in a negative way. Areas such as frustration with the system, cheating, poor instructional technique, low retention of information, cognitive overload, and poor leader/peer perceptions were identified through soldier participant interviews. Motivational issues dealing with negative feelings of relevancy and boredom with the instruction were also acknowledged. Additionally, difficulty in accessing the SSD system by soldiers, and over assumptions of soldiers’ levels of self-directed learning were also identified. This research contributes to the ongoing research needed dealing with soldier improvement through online learning.

People who have been diagnosed HIV positive often experience distress and anxiety due to uncertainties pertaining to the implications of an HIV positive status. These individuals are often reluctant to seek counselling and treatment due to the fear of being rejected and discriminated against (Parker, et al., 2002). There are limited formal networks for HIV support and psychological help in the South African context. Considering this, structured support groups were implemented for recently diagnosed HIV positive pregnant women. These women were recruited from ante natal clinics in Atteridgeville and Mamelodi as part of the Serithi project. Six support groups were implemented and facilitated by various experts including Masters students, of whom the researcher was part. This project is part of the larger study of the Serithi project in which interviews were conducted with three hundred and seventeen HIV positive pregnant women from disadvantaged locations of Tshwane. Based on these interviews, a support group intervention was developed. This research forms part of the evaluation of the support group intervention. The aim of this study was to explore the experiences of women who attended the support groups. Women who had attended 7-10 sessions were selected and interviewed individually using semi-structured interviews. With the permission of the participants, the discussions were tape recorded and transcribed. The data was analyzed, using qualitative research methods, from an interpretative phenomenological approach. This involved systematically studying meanings, themes and general descriptions of experiences by the research participants. The main findings in this study showed that women who participated in support groups adopted positive coping and behaviour that is conducive to their livelihood, learned more about HIV and AIDS, seem to have a positive future outlook and are overall empowered. These findings support previous research and literature in regards to the importance of social support in the form of support groups in effectively assisting HIV positive women in their journey to adjust to psychosocial consequence of the disease. / Dissertation (MA)--University of Pretoria, 2012. / Psychology / unrestricted