A real empirical study was conducted to illustrate the application of the unconstrained approach. The genuine study focused on the interaction effect of music self-concept and music-domain importance on the global self-concept. The result showed that structural equation analysis was advantageous over the traditional regression analysis, while such superiority of the structural equation modeling approach was more prominent in higher-order structural equation models. As expected, the estimated main and interaction effects in the quasi-standardized solution obtained using the centered data coincided with those using the standardized data in different order structural models. / Concepts and issues related to standardized solutions for the latent interaction model were discussed. Quasi-standardized solution was proposed and formulated by using the estimates of the original solution and the ordinary standardized solution. Some properties of the quasi-standardized solution were mathematically derived and proved, which included the demonstration that the main and interaction effects were scale free, so were the loading and the Chi-square of model fit, while t statistics of main and interaction effects were approximate scale free. / Six simulation studies, four for the latent interaction models, and two for the latent quadratic models were conducted to compare the performances of the four approaches. Results generally showed that the QML approach and the constrained approach behaved similarly, while the performance of the unconstrained approach was close to that of the GAPI approach. Under the normal distribution condition, the QML approach performed the best among the four approaches in terms of lack of bias, precision, and power. However, with moderate and large sample sizes (N=200 or above), the differences among the four approaches were systematically smaller, with similar bias and precision. Under nonnormal conditions, the unconstrained approach was more robust, with a smaller bias and predictable type I error rate (near the significant level), and its precision and power increased as the sample size increased. These results supported the use of the unconstrained approach for the analyses of latent interaction and quadratic models. / The unconstrained approach was extended to estimate interaction effects in latent growth models. With the indicators of the interaction term formed by the products of differences (rather than using the usual indicator product strategy), a simplified full interaction model for the latent growth model was proposed. The model was further simplified when only the interaction between change rates was considered. Importantly, the unconstrained approach was an appropriate method for analyses of such simplified full interaction model for latent growth model, which also constituted a unique contribution of this dissertation. / Through a series of related studies, the research attempted to identify better estimation approaches and modeling techniques for latent interaction and quadratic effects. The literature review provided a conceptual framework for the unconstrained approach which was recommended for its simplicity and robustness. Three other approaches, namely, the constrained, the partially constrained (i.e., the generalized appended product indicator, GAPI), and the quasi-maximum likelihood (QML) approaches were selected and compared with the unconstrained approach. / 溫忠粦. / 論文(哲學博士)--香港中文大學, 2005. / 參考文獻(p. 191-203). / Adviser: Kit-Tai Hau. / Source: Dissertation Abstracts International, Volume: 67-01, Section: A, page: 0160. / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts also in English. / School code: 1307. / Lun wen (Zhe xue bo shi)--Xianggang Zhong wen da xue, 2005. / Can kao wen xian (p. 191-203). / Wen Zhonglin.

Master of Science / Department of Computer Science / Cornelia Caragea / Doina Caragea / Social media has changed the way people communicate information. It has been noted that social media platforms like Twitter are increasingly being used by people and authorities in the wake of natural disasters. The year 2017 was a historic year for the USA in terms of natural calamities and associated costs. According to NOAA (National Oceanic and Atmospheric Administration), during 2017, USA experienced 16 separate billion-dollar disaster events, including three tropical cyclones, eight severe storms, two inland floods, a crop freeze, drought, and wild re. During natural disasters, due to the collapse of infrastructure and telecommunication, often it is hard to reach out to people in need or to determine what areas are affected. In such situations, Twitter can be a lifesaving tool for local government and search and rescue agencies. Using Twitter streaming API service, disaster-related tweets can be collected and analyzed in real-time. Although tweets received from Twitter can be sparse, noisy and ambiguous, some may contain useful information with respect to situational awareness. For example, some tweets express emotions, such as grief, anguish, or call for help, other tweets provide information specific to a region, place or person, while others simply help spread information from news or environmental agencies. To extract information useful for disaster response teams from tweets, disaster tweets need to be cleaned and classified into various categories. Topic modeling can help identify topics from the collection of such disaster tweets. Subsequently, a topic (or a set of topics) will be associated with a tweet. Thus, in this report, we will use Latent Dirichlet Allocation (LDA) to accomplish topic modeling for disaster tweets dataset.

One of the fundamental tasks in science is to find explainable relationships between observed phenomena. Recent work has addressed this problem by attempting to learn the structure of graphical models - especially Gaussian models - by the imposition of sparsity constraints. The graphical lasso is a popular method for learning the structure of a Gaussian model. It uses regularisation to impose sparsity. In real-world problems, there may be latent variables that confound the relationships between the observed variables. Ignoring these latents, and imposing sparsity in the space of the visibles, may lead to the pruning of important structural relationships. We address this problem by introducing an expectation maximisation (EM) method for learning a Gaussian model that is sparse in the joint space of visible and latent variables. By extending this to a conditional mixture, we introduce multiple structures, and allow side information to be used to predict which structure is most appropriate for each data point. Finally, we handle non-Gaussian data by extending each sparse latent Gaussian to a Gaussian copula. We train these models on a financial data set; we find the structures to be interpretable, and the new models to perform better than their existing competitors. A potential problem with the mixture model is that it does not require the structure to persist in time, whereas this may be expected in practice. So we construct an input-output HMM with sparse Gaussian emissions. But the main result is that, provided the side information is rich enough, the temporal component of the model provides little benefit, and reduces efficiency considerably. The GWishart distribution may be used as the basis for a Bayesian approach to learning a sparse Gaussian. However, sampling from this distribution often limits the efficiency of inference in these models. We make a small change to the state-of-the-art block Gibbs sampler to improve its efficiency. We then introduce a Hamiltonian Monte Carlo sampler that is much more efficient than block Gibbs, especially in high dimensions. We use these samplers to compare a Bayesian approach to learning a sparse Gaussian with the (non-Bayesian) graphical lasso. We find that, even when limited to the same time budget, the Bayesian method can perform better. In summary, this thesis introduces practically useful advances in structure learning for Gaussian graphical models and their extensions. The contributions include the addition of latent variables, a non-Gaussian extension, (temporal) conditional mixtures, and methods for efficient inference in a Bayesian formulation.

PhD / This thesis describes an interview study of forty five professionally accomplished male and female designers and architects. The study considers how each respondent designs and makes discoveries throughout conceptual design. How they start designing, what they attempt to achieve, the means they employ, how they cope with getting stuck, their breakthroughs and discoveries and the circumstances of these experiences, are the main ingredients of the study. The aim of the research is to estimate the extent to which designing may be regarded as an insightful activity, by investigating experiences of discoveries as reported by the respondents. Throughout the thesis, discoveries or ideas occurring to respondents when they are not actively designing, an apparent outcome of a latent designing or preparation activity, are referred to as cold discoveries. This label is used to distinguish these discoveries from discoveries that emerge in the run of play, when individuals are actively designing. The latter are referred to as hot discoveries. The relative insightfulness of hot and cold discoveries is also investigated. In general, the evidence from the research suggests that designing is significantly insightful. Most respondents (39:45) reported experiences of insights that have contributed to their designing. In addition there is strong evidence that cold discoveries are considerably more important, both quantitatively and qualitatively, than is currently recognized. More than half of the respondents (25:45) reported the experience of cold discoveries, many after disengaging from designing, when they had been stuck. Being stuck means they were experiencing frustration, or had recognised they were not making satisfactory progress in attempts to resolve some aspect of conceptual design. Typically these respondents reported experiencing discoveries while doing other work, performing some physical activity, resting, or very soon after resuming work. They had elected to let ideas come to them, rather than persist in searching and this strategy was successful. Moreover, many respondents (10:45) described positive attributes of cold discoveries using terms such as stronger, more potent, or pushes boundaries, which suggest their cold discoveries are more insightful than their hot discoveries. Many respondents associated their cold discoveries with mental activities such as incubation, a concept identified by Gestalt theorists nearly a century ago. They used a range of informal terms, such as ideas ticking over, or percolating away. These apparently uncontrolled mental experiences, which I refer to generically as latent preparation, varied from one respondent to another in when, where and how they occurred. Latent preparation or its outcomes, in the form of interruptive thoughts, apparently takes place at any time and during different states of consciousness and attentiveness. It appears to be, at different times, unplanned, unintentional, undirected, unnoticed, or unconscious, in combinations, not necessarily all at once. It is clearly not only an unconscious process. This suggests one, or more of the following; 1) that incubation is only a component of latent preparation, or 2) that the conventional view of incubation, as an unconscious process, does not adequately account for the range of insightful experiences of mentally productive people, such as designers, or 3) that the old issue of whether incubation is a conscious, or an unconscious process, is not vital to a systematic investigation of insightful discovery. The thesis concludes by considering prospects for further research and how the research outcomes could influence education. Apart from the findings already described, statements by the respondents about personal attributes, designing, coping with being stuck and discoveries, were wide ranging, resourceful and down-to-earth, suggesting there are many ways for individuals to become proficient, creative designers at the high end of their profession. A major implication for future research is that latent preparation may be found as readily among highly motivated and skilled individuals in other occupations unrelated to architecture or designing. The evidence of the research so far suggests there is much to be learned about latent preparation that can be usefully applied, for the benefit of individuals aiming to be designers, or simply wanting to become more adept at intervening, transforming and managing unexpected and novel situations of any kind.

As a new topic in text mining, novelty detection is a natural extension of information retrieval systems, or search engines. Aiming at refining raw search results by filtering out old news and saving only the novel messages, it saves modern people from the nightmare of information overload. One of the difficulties in novelty detection is the inherent ambiguity of language, which is the carrier of information. Among the sources of ambiguity, synonymy proves to be a notable factor. To address this issue, previous studies mainly employed WordNet, a lexical database which can be perceived as a thesaurus. Rather than borrowing a dictionary, we proposed a statistical approach employing Latent Semantic Indexing (LSI) to learn semantic relationship automatically with the help of language resources. To apply LSI which involves matrix factorization, an immediate problem is that the dataset in novelty detection is dynamic and changing constantly. As an imitation of real-world scenario, texts are ranked in chronological order and examined one by one. Each text is only compared with those having appeared earlier, while later ones remain unknown. As a result, the data matrix starts as a one-row vector representing the first report, and has a new row added at the bottom every time we read a new document. Such a changing dataset makes it hard to employ matrix methods directly. Although LSI has long been acknowledged as an effective text mining method when considering semantic structure, it has never been used in novelty detection, nor have other statistical treatments. We tried to change this situation by introducing external text source to build the latent semantic space, onto which the incoming news vectors were projected. We used the Reuters-21578 dataset and the TREC data as sources of latent semantic information. Topics were divided into years and types in order to take the differences between them into account. Results showed that LSI, though very effective in traditional information retrieval tasks, had only a slight improvement to the performances for some data types. The extent of improvement depended on the similarity between news data and external information. A probing into the co-occurrence matrix attributed such a limited performance to the unique features of microblogs. Their short sentence lengths and restricted dictionary made it very hard to recover and exploit latent semantic information via traditional data structure.

The Uses of Objects Task is a widely used assessment of creative performance, but it relies on subjective scoring methods for evaluation. A new version of the task was devised using Latent Semantic Analysis (LSA), a computational tool used to measure semantic distance. 135 participants provided as many creative uses for as they could for 20 separate objects. Responses were analyzed for strategy use, category switching, variety, and originality of responses, as well as subjective measure of creativity by independent raters. The LSA originality measure was more reliable than the subjective measure, and values averaged over participants correlated with both subjective evaluations and self-assessment of creativity. The score appeared to successfully isolate the creativity of the people themselves, rather than the potential creativity afforded by a given object.

The Uses of Objects Task is a widely used assessment of creative performance, but it relies on subjective scoring methods for evaluation. A new version of the task was devised using Latent Semantic Analysis (LSA), a computational tool used to measure semantic distance. 135 participants provided as many creative uses for as they could for 20 separate objects. Responses were analyzed for strategy use, category switching, variety, and originality of responses, as well as subjective measure of creativity by independent raters. The LSA originality measure was more reliable than the subjective measure, and values averaged over participants correlated with both subjective evaluations and self-assessment of creativity. The score appeared to successfully isolate the creativity of the people themselves, rather than the potential creativity afforded by a given object.

Understanding the neural mechanisms underlying the extinction of maladaptive behaviors has become increasingly relevant. Extinction, or the reduction of a response due to lack of reinforcement, is believed to be “new learning.” Most extinction paradigms involve the performance of the previously reinforced response in the absence of reinforcement in order for extinction to occur. Conversely, latent extinction is a cognitive form of learning in which the previously rewarded response is not made during extinction training. However, until now the neurobiological basis of latent extinction has remained unknown. This dissertation has three aims to examine the neurobiological basis of latent extinction. Previous research has shown latent extinction to be impaired following hippocampal inactivation and the goal of Aim 1 was to examine other neural systems potentially involved in latent extinction through examination of brain structures such as the dorsal striatum, medial prefrontal cortex, and basolateral amygdala. Additionally, the neurochemical basis of latent extinction is unidentified; therefore Aim 2 addressed this question, specifically investigating the glutamatergic system through both NMDA receptor agonism and antagonism. Finally, understanding latent extinction may be useful for the extinction of drug addiction. Aim 3 was to examine some clinical implications for the extinction of drug addiction utilizing latent extinction following maze running for an oral cocaine reward. Reversible neural inactivation studies using the sodium channel blocker bupivacaine demonstrated a selective impairment of response extinction following dorsal striatum inactivation, but no effect on either latent or response extinction following medial prefrontal cortex or basolateral amygdala inactivation. These results, coupled with previous data from our lab demonstrate a double dissociation for extinction behavior. Further, peripheral NMDA receptor agonism with D-cyloserine enhances latent extinction and intra-hippocampal NMDA receptor antagonism with AP5 impairs latent extinction, identifying a role for the glutamatergic system in latent extinction. Finally, oral cocaine administration during acquisition selectively impairs latent extinction indicating that drug use affects the relive use of multiple memory systems during extinction. Overall, the multiple memory systems theory and latent extinction provide a framework with which to further understand the neural mechanisms of extinction behavior.

Text categorization to automatically assign documents into the appropriate pre-defined category or categories is essential to facilitating the retrieval of desired documents efficiently and effectively from a huge text depository, e.g., the world-wide web. Most techniques, however, suffer from the feature selection problem and the vocabulary mismatch problem. A few research works have addressed on text categorization via text summarization to reduce the size of documents, and consequently the number of features to consider, while some proposed using latent semantic indexing (LSI) to reveal the true meaning of a term via its association with other terms. Few works, however, have studied the joint effect of text summarization and the semantic dimension reduction technique in the literature. The objective of this research is thus to propose a practical approach, SBDR to deal with the above difficulties in text categorization tasks. Two experiments are conducted to validate our proposed approach. In the first experiment, the results show that text summarization does improve the performance in categorization. In addition, to construct important sentences, the association terms of both noun-noun and noun-verb pairs should be considered. Results of the second experiment indicate slight better performance with the approach of adopting LSI exclusively (i.e. no summarization) than that with SBDR (i.e. with summarization). Nonetheless, the minor accuracy reduction can be largely compensated for the computational time saved using LSI with text summarized. The feasibility of the SBDR approach is thus justified.

Thesis (M.S.)--California Polytechnic State University, 2009. / Mode of access: Internet. Title from PDF title page; viewed on Jan. 6, 2010. Major professor: Dr. Michael Haungs. "Presented to the faculty of California Polytechnic State University, San Luis Obispo." "In partial fulfillment of the requirements for the degree [of] Master of Science in Computer Science." "Aug 2009." Includes bibliographical references (p. 76-78).