Matching Items (185)
Filtering by
- All Subjects: Machine Learning
- Genre: Doctoral Dissertation
- Resource Type: Text
Description
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.
ContributorsYuan, Lei (Author) / Ye, Jieping (Thesis advisor) / Wang, Yalin (Committee member) / Xue, Guoliang (Committee member) / Kumar, Sudhir (Committee member) / Arizona State University (Publisher)
Created2013
Description
This work presents two complementary studies that propose heuristic methods to capture characteristics of data using the ensemble learning method of random forest. The first study is motivated by the problem in education of determining teacher effectiveness in student achievement. Value-added models (VAMs), constructed as linear mixed models, use students’ test scores as outcome variables and teachers’ contributions as random effects to ascribe changes in student performance to the teachers who have taught them. The VAMs teacher score is the empirical best linear unbiased predictor (EBLUP). This approach is limited by the adequacy of the assumed model specification with respect to the unknown underlying model. In that regard, this study proposes alternative ways to rank teacher effects that are not dependent on a given model by introducing two variable importance measures (VIMs), the node-proportion and the covariate-proportion. These VIMs are novel because they take into account the final configuration of the terminal nodes in the constitutive trees in a random forest. In a simulation study, under a variety of conditions, true rankings of teacher effects are compared with estimated rankings obtained using three sources: the newly proposed VIMs, existing VIMs, and EBLUPs from the assumed linear model specification. The newly proposed VIMs outperform all others in various scenarios where the model was misspecified. The second study develops two novel interaction measures. These measures could be used within but are not restricted to the VAM framework. The distribution-based measure is constructed to identify interactions in a general setting where a model specification is not assumed in advance. In turn, the mean-based measure is built to estimate interactions when the model specification is assumed to be linear. Both measures are unique in their construction; they take into account not only the outcome values, but also the internal structure of the trees in a random forest. In a separate simulation study, under a variety of conditions, the proposed measures are found to identify and estimate second-order interactions.
ContributorsValdivia, Arturo (Author) / Eubank, Randall (Thesis advisor) / Young, Dennis (Committee member) / Reiser, Mark R. (Committee member) / Kao, Ming-Hung (Committee member) / Broatch, Jennifer (Committee member) / Arizona State University (Publisher)
Created2013
Description
Consumers search before making virtually any purchase. The notion that consumers engage in costly search is well-understood to have deep implications for market performance. However to date, no theoretical model allows for the observation that consumers often purchase more than a single product in an individual shopping occasion. Clothing, food, books, and music are but four important examples of goods that are purchased many items at a time. I develop a modeling approach that accounts for multi-purchase occasions in a structural way. My model shows that as preference for variety increases, so does the size of the consideration set. Search models that ignore preference for variety are, therefore, likely to under-predict the number of products searched. It is generally thought that lower search costs increase retail competition which pushes prices and assortments down. However, I show that there is an optimal number of products to offer depending on the intensity of consumer search costs. Consumers with high search costs prefer to shop at a store with a large assortment of goods and purchase multiple products, even if the prices that firm charges is higher than competing firms' prices. On the other hand, consumers with low search costs tend to purchase fewer goods and shop at the stores that have lower prices, as long as the store has a reasonable assortment offering. The implications for market performance are dramatic and pervasive. In particular, the misspecification of demand model in which search is important and/or multiple discreteness is observed will produce biased parameter estimates leading to erroneous managerial conclusions.
ContributorsAllender, William Jacob (Author) / Richards, Timothy J. (Thesis advisor) / Park, Sungho (Committee member) / Hamilton, Stephen (Committee member) / Arizona State University (Publisher)
Created2013
Description
There have been multiple calls for research on consumers' responses to social issues, regulatory changes, and corporate behavior. Thus, this dissertation proposes and tests a conceptual framework of parents' responses to government regulations and corporate social responsibility (CSR) that address juvenile obesity. This research builds on Attribution Theory to examine the impact of government regulations and CSR on consumers' attitudes and their subsequent behavior. Three pilot studies and three main experiments were conducted; a between-subjects and randomized experimental design being used to capture the effects of regulations and corporate actions on product satisfaction, company evaluations, and behavioral intentions, while examining the mediating role of attributions of responsibility for a negative product outcome. This research has implications for policy makers and marketing practitioners and scholars. This is the first study to offer a new perspective, based on attributions of blame, to explain the mechanism that drives consumers' responses to government regulations. Considering numerous calls for government actions that address childhood obesity, it is important to understand how and why consumers respond to such regulations. The results illustrated that certain policies may have unintended consequences due to unexpected attributions of blame for unhealthy products. Only recently have researchers tried to address the psychological mechanism through which CSR has an impact on consumers' attitudes and behavior. To date, few studies have investigated attributions as a mediating variable in the transfer of CSR associations on consumer responses. Nonetheless, this is the first study that concentrates on attributions of responsibility, per se, to explain the impact of CSR on company evaluations. This dissertation extends previous research, where locus, stability, and controllability mediated the relationship between CSR and attributions of blame; the degree of blame being consequential to brand evaluations. The current results suggest that attributions of responsibility, per se, mediate the impact of CSR on company evaluations. Additionally, attributions of blame are measured as the degree to which consumers take personal responsibility for a negative product outcome. This highlights a new role of the CSR construct, as a moderator of consumers' self-serving bias, a fundamental psychological response that has been neglected in the marketing literature.
ContributorsDumitrescu, Claudia (Author) / Shaw Hughner, Renée (Thesis advisor) / Schmitz, Troy G. (Committee member) / Seperich, George (Committee member) / Shultz, Ii, Clifford J. (Committee member) / Arizona State University (Publisher)
Created2013
Description
Image understanding has been playing an increasingly crucial role in vision applications. Sparse models form an important component in image understanding, since the statistics of natural images reveal the presence of sparse structure. Sparse methods lead to parsimonious models, in addition to being efficient for large scale learning. In sparse modeling, data is represented as a sparse linear combination of atoms from a "dictionary" matrix. This dissertation focuses on understanding different aspects of sparse learning, thereby enhancing the use of sparse methods by incorporating tools from machine learning. With the growing need to adapt models for large scale data, it is important to design dictionaries that can model the entire data space and not just the samples considered. By exploiting the relation of dictionary learning to 1-D subspace clustering, a multilevel dictionary learning algorithm is developed, and it is shown to outperform conventional sparse models in compressed recovery, and image denoising. Theoretical aspects of learning such as algorithmic stability and generalization are considered, and ensemble learning is incorporated for effective large scale learning. In addition to building strategies for efficiently implementing 1-D subspace clustering, a discriminative clustering approach is designed to estimate the unknown mixing process in blind source separation. By exploiting the non-linear relation between the image descriptors, and allowing the use of multiple features, sparse methods can be made more effective in recognition problems. The idea of multiple kernel sparse representations is developed, and algorithms for learning dictionaries in the feature space are presented. Using object recognition experiments on standard datasets it is shown that the proposed approaches outperform other sparse coding-based recognition frameworks. Furthermore, a segmentation technique based on multiple kernel sparse representations is developed, and successfully applied for automated brain tumor identification. Using sparse codes to define the relation between data samples can lead to a more robust graph embedding for unsupervised clustering. By performing discriminative embedding using sparse coding-based graphs, an algorithm for measuring the glomerular number in kidney MRI images is developed. Finally, approaches to build dictionaries for local sparse coding of image descriptors are presented, and applied to object recognition and image retrieval.
ContributorsJayaraman Thiagarajan, Jayaraman (Author) / Spanias, Andreas (Thesis advisor) / Frakes, David (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Customer participation in service conversations: an investigation of the dynamics of service context
Description
In this research, I focus on service conversations in professional services. For most Business-to-Business or Business-to-Consumer professional services, the service conversation is an important part of the service experience and is critical to solutions co-creation as well as customer satisfaction. In this research, I examine service conversation sequences at the micro-level and explore two important research questions: (1) how do I explain the dynamics of moment-by-moment Customer Participation in Service Conversations (CPSC)? and (2) how do the temporal and process dynamics of CPSC relate to customer satisfaction and solution compliance? From a dynamic context perspective, I develop a conceptual framework that explains the co-existence of stable and dynamic customer participation behavior in a service conversation. I conduct a series of lab experiments and an observation study of online conversations between 173 customers and 52 doctors to empirically validate the conceptual framework. This research demonstrates that at any given moment, customers manage their information sharing and interaction control based on their mental representation of the context complexity. Although the context-behavior relationships are stable, the service conversation context is dynamic. The service provider's behavior can constantly change and introduce new context cues. When the context changes so does the CPSC behavior. Finally, this research shows that to improve customer satisfaction, customer perceived service quality, and customer solution compliance, service providers should focus on helping customers reduce their perceived context complexity as early as possible, by providing information and educating customers. This research makes important theoretical and managerial contributions. Theoretically, it defines and develops measures of service context complexity in terms of its psychological features. It develops a conceptual framework to explain the temporal dynamics of CPSC on multi-dimensions. Empirically, this research adopts a phase-based sequence analysis approach and uses a negative bi-nominal model to examine the temporal process effect of the service conversation on service outcomes. Managerially, the research findings provide firms important and actionable guidelines to manage conversation-based professional services.
ContributorsWang, Si (Author) / Binter, Mary Jo (Thesis advisor) / Ostrom, Amy L. (Committee member) / Olsen, G. Douglas (Committee member) / Arizona State University (Publisher)
Created2013
Description
With the increase in computing power and availability of data, there has never been a greater need to understand data and make decisions from it. Traditional statistical techniques may not be adequate to handle the size of today's data or the complexities of the information hidden within the data. Thus knowledge discovery by machine learning techniques is necessary if we want to better understand information from data. In this dissertation, we explore the topics of asymmetric loss and asymmetric data in machine learning and propose new algorithms as solutions to some of the problems in these topics. We also studied variable selection of matched data sets and proposed a solution when there is non-linearity in the matched data. The research is divided into three parts. The first part addresses the problem of asymmetric loss. A proposed asymmetric support vector machine (aSVM) is used to predict specific classes with high accuracy. aSVM was shown to produce higher precision than a regular SVM. The second part addresses asymmetric data sets where variables are only predictive for a subset of the predictor classes. Asymmetric Random Forest (ARF) was proposed to detect these kinds of variables. The third part explores variable selection for matched data sets. Matched Random Forest (MRF) was proposed to find variables that are able to distinguish case and control without the restrictions that exists in linear models. MRF detects variables that are able to distinguish case and control even in the presence of interaction and qualitative variables.
ContributorsKoh, Derek (Author) / Runger, George C. (Thesis advisor) / Wu, Tong (Committee member) / Pan, Rong (Committee member) / Cesta, John (Committee member) / Arizona State University (Publisher)
Created2013
Description
Effective modeling of high dimensional data is crucial in information processing and machine learning. Classical subspace methods have been very effective in such applications. However, over the past few decades, there has been considerable research towards the development of new modeling paradigms that go beyond subspace methods. This dissertation focuses on the study of sparse models and their interplay with modern machine learning techniques such as manifold, ensemble and graph-based methods, along with their applications in image analysis and recovery. By considering graph relations between data samples while learning sparse models, graph-embedded codes can be obtained for use in unsupervised, supervised and semi-supervised problems. Using experiments on standard datasets, it is demonstrated that the codes obtained from the proposed methods outperform several baseline algorithms. In order to facilitate sparse learning with large scale data, the paradigm of ensemble sparse coding is proposed, and different strategies for constructing weak base models are developed. Experiments with image recovery and clustering demonstrate that these ensemble models perform better when compared to conventional sparse coding frameworks. When examples from the data manifold are available, manifold constraints can be incorporated with sparse models and two approaches are proposed to combine sparse coding with manifold projection. The improved performance of the proposed techniques in comparison to sparse coding approaches is demonstrated using several image recovery experiments. In addition to these approaches, it might be required in some applications to combine multiple sparse models with different regularizations. In particular, combining an unconstrained sparse model with non-negative sparse coding is important in image analysis, and it poses several algorithmic and theoretical challenges. A convex and an efficient greedy algorithm for recovering combined representations are proposed. Theoretical guarantees on sparsity thresholds for exact recovery using these algorithms are derived and recovery performance is also demonstrated using simulations on synthetic data. Finally, the problem of non-linear compressive sensing, where the measurement process is carried out in feature space obtained using non-linear transformations, is considered. An optimized non-linear measurement system is proposed, and improvements in recovery performance are demonstrated in comparison to using random measurements as well as optimized linear measurements.
ContributorsNatesan Ramamurthy, Karthikeyan (Author) / Spanias, Andreas (Thesis advisor) / Tsakalis, Konstantinos (Committee member) / Karam, Lina (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
Unrestricted Mexican exports of sugar into the U.S. is considered the most pressing issue facing the U.S. sugar industry. The goal of this dissertation is to analyze the trade of sugar between Mexico and the U.S. as well as analyze additional primary issues confronting the U.S. sugar industry. Chapters 1 and 2 provide an introduction to the U.S. sugar industry. Chapters 3 through 6 develop trade models which analyze sugar trade between Mexico and the U.S. The trade models estimate how NAFTA, USDA sugar forecast errors and Mexican ownership of twenty percent of the Mexican sugar industry each impact U.S. producer surplus and Mexican welfare. Results validate that U.S. producer surplus and in some instances Mexican welfare were decreased by full implementation of NAFTA. U.S. producer surplus and Mexican welfare were decreased due to USDA sugar production forecasting errors. U.S. producer surplus would be increased if the Mexican government did not own twenty percent of Mexican sugar production. Using an online choice experiment, Chapter 7 assesses U.S. consumers' preferences and willingness to pay (WTP) for imported and genetically modified (GM) labeled sugar and sugar in soft drinks. Results indicate that consumers prefer bags of sugar and soft drinks labeled as "Not GM". Furthermore, consumers prefer sugar from Canada and the U.S. over sugar from Mexico, Brazil and the Philippines. Evidence is also provided that participants are more likely to choose actual products in the choice set rather than the "none of these" options when controlling for hypothetical bias by using consequentiality techniques. A non-hypothetical experimental auction was used in Chapter 8 to determine consumers' WTP for soft drinks labeled with sweetener and calorie information and analyzed the role of taste panels in an experimental auction. Results indicate that sugar is consumers' most preferred sweetener and calorie labeling is ineffective at influencing consumers to choose healthier soft drinks. Including taste in an experimental auction caused significant reductions in consumers' WTP for all soft drinks. Chapter 9 concludes by summarizing the results of this dissertation and discussing the future challenges facing the U.S. sugar industry.
ContributorsLewis, Karen Elizabeth (Author) / Schmitz, Troy (Thesis advisor) / Grebitus, Carola (Committee member) / Manfredo, Mark (Committee member) / Ketcham, Andrea (Committee member) / Arizona State University (Publisher)
Created2014
Description
Identifying factors associated with service infusion success has become an important issue in theory and practice, as manufacturers turn to services to advance performance. The goals of this dissertation are to identify the key factors associated with service infusion success and develop an integrative framework and associated research propositions to isolate the underlying determinants of successful hybrid solution strategies for business customers. This dissertation is comprised of two phases. The first phase taps into the experience and learning gained by executives from Fortune-100 manufacturing firms who are managing the transition from goods to hybrid offerings for their customers. A discovery-oriented, theory-in-use approach is adopted to glean insights concerning the factors that facilitate and hinder those service transition strategies. Twenty-eight interviews were conducted with key executives, transcripts were analyzed and key themes were identified with special attention directed to the particular capabilities that managers consider crucial for successful service-growth strategies. One such capability centers on the ability of a firm to successfully transfer newly-developed hybrid solutions from one customer engagement to another. Building on this foundation, phase two involves a case study that provides an in-depth examination of the hybrid offering replication process in a business-to-business firm attempting to replicate four strategic hybrid offerings. Emergent themes, based on 13 manager interviews, reveal factors that promote or impede successful hybrid offering transfer. Among the factors that underlie successful hybrid offering transfers across customer engagements are close customer relationships, a clear value proposition embraced by organizational numbers, an accurate forecast of market potential, and collaborative working relationships across units. The findings from the field studies provided a catalyst for a deeper examination of existing literature and formed the building blocks for the conceptual model and several key research propositions related to the successful transfer of hybrid offerings. The model isolates five sets of factors that influence the hybrid offering transfer process, including the characteristics of (1) the source project team, (2) the seeking project team, (3) the hybrid offering, (4) the relationship exchange, and (5) the customer. The conceptualization isolates the critical role that the customer assumes in service infusion strategy implementation.
ContributorsSalas, Jim (Author) / Walker, Beth (Thesis advisor) / Hutt, Michael D. (Thesis advisor) / Park, Sungho (Committee member) / Ulaga, Wolfgang (Committee member) / Arizona State University (Publisher)
Created2013