Theses and Dissertations from UMD
Permanent URI for this communityhttp://hdl.handle.net/1903/2
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a give thesis/dissertation in DRUM
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
7 results
Search Results
Item Structured discovery in graphs: Recommender systems and temporal graph analysis(2024) Peyman, Sheyda Do'a; Lyzinski, Vince V.; Applied Mathematics and Scientific Computation; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Graph-valued data arises in numerous diverse scientific fields ranging from sociology, epidemiology and genomics to neuroscience and economics.For example, sociologists have used graphs to examine the roles of user attributes (gender, class, year) at American colleges and universities through the study of Facebook friendship networks and have studied segregation and homophily in social networks; epidemiologists have recently modeled Human-nCov protein-protein interactions via graphs, and neuroscientists have used graphs to model neuronal connectomes. The structure of graphs, including latent features, relationships between the vertex and importance of each vertex are all highly important graph properties that are main aspects of graph analysis/inference. While it is common to imbue nodes and/or edges with implicitly observed numeric or qualitative features, in this work we will consider latent network features that must be estimated from the network topology.The main focus of this text is to find ways of extracting the latent structure in the presence of network anomalies. These anomalies occur in different scenarios: including cases when the graph is subject to an adversarial attack and the anomaly is inhibiting inference, and in the scenario when detecting the anomaly is the key inference task. The former case is explored in the context of vertex nomination information retrieval, where we consider both analytic methods for countering the adversarial noise and also the addition of a user-in-the-loop in the retrieval algorithm to counter potential adversarial noise. In the latter case we use graph embedding methods to discover sequential anomalies in network time series.Item Spatiotemporal Dynamics and Functional Organization of Auditory Cortex Networks(2021) Bowen, Zac; Kanold, Patrick O; Losert, Wolfgang; Biophysics (BIPH); Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The sensory cortices of the brain are highly complex systems that are uniquely adapted to reliably process any encountered sensory stimulus. Sensory stimuli such as sound are encoded in large populations of neurons that exhibit some functional organization in the cortex. For example, the auditory cortex has a characteristic organization of sound frequency by which neuronal responses are organized. However, this organization is a broad approximation of more complex and diverse functional properties of individual neurons. Furthermore, on a finer temporal scale, the moment-to-moment activity dynamics of populations of neurons are incredibly complex. Numerous studies have shown that spatiotemporal cascades of co-active neurons organize as neuronal avalanches possessing certain characteristics such as size, duration, and shape that fit the parameters of a critical system. Nevertheless, it remains that the exact manner in which neuronal populations encode information is still not fully understood. This dissertation makes use of neuroimaging data acquired with 2-photon calcium imaging of the auditory cortex in awake mice to investigate the spatiotemporal and functional organization of active neuronal populations in auditory cortex at a range of temporal and spatial scales. I aimed to gain a deeper understanding into how neuronal population dynamics and the underlying network organization contribute to sound encoding in auditory cortex. I studied input and associative layers of auditory cortex (L4 and L2/3) in a mouse model with normal hearing and another with age-related hearing loss due to loss of proper cochlear function to high-frequency sound. L4 and L2/3 contained populations of neurons with a large diversity in functional properties, though diversity was reduced in the hearing loss model due to paucity of high frequency tuned neurons. Despite the diverse tuning in both, similarly responding neurons tended to be co-localized in cortical space. I found that this result extended to volumetric samples of L2/3 where large populations of neurons contained a functional network architecture indicative of small-world topology. Furthermore, I demonstrated that L4 and L2/3 contain ensembles of co-active neurons indicative of critical dynamics in both the absence and presence of a stimulus. Finally, I developed software that facilitates real-time quantification of neuronal populations during an experiment which opens the door for novel closed-loop experiment design. This dissertation provides several avenues for further investigation into neuronal population coding and dynamics, functional network topology, and provides the groundwork for closed-loop experimental design.Item Financial Reporting: A Look At Different Settings(2013) Felix, Robert; Cheng, Dr. Shijun; Business and Management: Accounting & Information Assurance; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The first of two essays examines whether financial reporting is influenced when a firm shares a director with a "central" firm. Central firms are those which are well-connected within the network of firms formed by shared board of directors. Centrality is a driver of influence and since social networks are a channel to spread information, central firms could transmit reporting practices. However, because financial reporting style is presumably firm specific, the central firm's reporting may not be effective for a focal firm. I examine the effect of central firm conservatism and discretionary accruals on the same focal firm attributes. The results show that focal firm conservatism is influenced by that of the central firm after the two firms become interlocked and that influence is concentrated in the first year. However, a firm adopted central firm discretionary accruals over a longer time horizon. The finding was robust to a variety of alternate explanations. Overall, the findings shed light on how financial reporting spreads through a network and adds to our understanding of how influence occurs between two interlocked firms. The second essay examines municipal reporting manipulation. Municipalities use fund accounting to separately track each activity in self-balancing set of accounts. I focus on the general fund, the largest fund, which uses governmental accounting, and the enterprise fund, which accounts for business-like operations and uses corporate-like accounting. Municipalities have a different organizational objective than corporations and could desire to report a small increase in the general fund bottom line to avoid taxpayer's backlash or they could wish to build up their fund balance to for future use. The enterprise fund incentives are also unclear. I find that operating transfers between funds (discretionary accruals) are used in the general (enterprise), but not the enterprise (general), fund to systematically manipulate its bottom line downward. Accordingly, each fund is manipulated downwards using a method that is in line with its accounting system. Further analysis shows that the general fund results are more pronounced in municipalities with heavy citizen involvement. The findings also highlight that institutional factors do not impact both funds in the same manner.Item NETWORK MODELS OF REGIONAL INNOVATION CLUSTERS AND THEIR IMPACT ON ECONOMIC GROWTH(2012) Dempwolf, Christopher Scott; Howland, Marie; Urban and Regional Planning and Design; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This research uses social network analysis to develop models of regional innovation clusters using data from patent applications and other sources. These new models are more detailed than current industry cluster models, and they reveal actual and potential relationships among firms that industry cluster models cannot. The network models can identify specific clusters of firms with high potential for manufacturing job growth where business retention and expansion efforts may be targeted. They can also identify dense clusters of talent where innovation and entrepreneurial efforts may be targeted. Finally, this research measures relationships between network structure at the time of patent application and manufacturing job growth in subsequent years. This will permit the translation of a wide range of network-building activities into the ubiquitous "jobs created" metric. These new tools will help economic developers focus resources on high-yield activities, and measure the results of networking activities more effectively. There are three parts to this research. First, it evaluates the uses of social network analysis (SNA) in planning, reviewing the literature and empirical research where SNA has been used in planning related studies. Second, it presents the construction if innovation network models, covering methodology, data, results and direct applications of the network models themselves. Models are constructed for Pennsylvania between 1990 and 2007. The methodology presents a significant innovation in how networks and geography are modeled, embedding counties in the network as place nodes. The resulting network models more accurately reflect the complex and multiple relationships that firms and inventors have with each other and the locations where they interact. This approach makes it possible to evaluate relationships between innovation and economic growth at a smaller geographic level (counties) than previous research. Third, this research presents an econometric model that evaluates the influence of network structure on county-level manufacturing employment and value added. Network structure is measured in the year of patent application, with manufacturing employment and value added being measured annually for each subsequent year. Differences in network structure generally reflect differences in the level of social capital embedded in different parts of the network. I find that network structure influences manufacturing employment within three years (longer for medical devices and pharmaceuticals) but does not influence value added.Item Computationally Comparing Biological Networks and Reconstructing Their Evolution(2012) Patro, Robert; Kingsford, Carleton L; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Biological networks, such as protein-protein interaction, regulatory, or metabolic networks, provide information about biological function, beyond what can be gleaned from sequence alone. Unfortunately, most computational problems associated with these networks are NP-hard. In this dissertation, we develop algorithms to tackle numerous fundamental problems in the study of biological networks. First, we present a system for classifying the binding affinity of peptides to a diverse array of immunoglobulin antibodies. Computational approaches to this problem are integral to virtual screening and modern drug discovery. Our system is based on an ensemble of support vector machines and exhibits state-of-the-art performance. It placed 1st in the 2010 DREAM5 competition. Second, we investigate the problem of biological network alignment. Aligning the biological networks of different species allows for the discovery of shared structures and conserved pathways. We introduce an original procedure for network alignment based on a novel topological node signature. The pairwise global alignments of biological networks produced by our procedure, when evaluated under multiple metrics, are both more accurate and more robust to noise than those of previous work. Next, we explore the problem of ancestral network reconstruction. Knowing the state of ancestral networks allows us to examine how biological pathways have evolved, and how pathways in extant species have diverged from that of their common ancestor. We describe a novel framework for representing the evolutionary histories of biological networks and present efficient algorithms for reconstructing either a single parsimonious evolutionary history, or an ensemble of near-optimal histories. Under multiple models of network evolution, our approaches are effective at inferring the ancestral network interactions. Additionally, the ensemble approach is robust to noisy input, and can be used to impute missing interactions in experimental data. Finally, we introduce a framework, GrowCode, for learning network growth models. While previous work focuses on developing growth models manually, or on procedures for learning parameters for existing models, GrowCode learns fundamentally new growth models that match target networks in a flexible and user-defined way. We show that models learned by GrowCode produce networks whose target properties match those of real-world networks more closely than existing models.Item Adapting to Norms at the United Nations: the Abortion-Rights and Anti-Abortion Networks(2007-11-20) Swinski, June Samuel; Conca, Ken; Government and Politics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation examines the practical effects of international norm construction for social movements attempting to navigate the UN system, specifically UN global conferences. Do norms become ingrained in the practices of intergovernmental organizations to such an extent that they hinder a movement with different norms or help a movement that conforms to them? In studying the UN and especially UN global conferences on issues of social significance, it has been argued that the norms stemming from classic Lockean liberalism, such as emphasis on individual liberties, a rights-based framework for developing policy, and progress through science and reason, are embodied in the procedures and frameworks of UN global conferences. I compare the strategies and influence of the abortion-rights and anti-abortion movements over time at the UN, particularly through the International Conferences on Population and Development, and trace how each movement has adjusted its strategies to accommodate the normative context it has encountered at the UN. I use a combined structural and agency-oriented framework that identifies the concrete mechanisms and processes through which the interplay of movement ideology and institutional-normative context may constrain or facilitate a social movement's actions within the UN system. What I've found in my research is that the abortion-rights network has had more success in actually influencing the debate and changing the language of population policy to reflect their goals, whereas the influence of the anti-abortion network can really only be measured by the language that they have blocked. But it is important to note that both the abortion-rights network and the anti-abortion network have adjusted over time to the UN in terms of their strategies, which is interesting because of the more progressive character of one, and the conservative character of the other. However, the progressive and conservative characters of the two movements still affected how easily each movement adapted to these norms at the UN, and the success of their strategies in that forum.Item Essays on the impact of social interactions on economic outcomes(2007-06-03) Perez Rojas, Nathalia; Kranton, Rachel; Economics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation consists of two essays, which address the question of how social interactions shape economic outcomes. The first essay examines crime and criminal networks. The second one studies immigration, assimilation, and ethnic enclaves. The first essay offers a formal model of crime. Criminals often do not act alone. Rather, they form networks of collaboration. How does law enforcement affect criminal activity and structure of those networks? Using a network game, I show that increased enforcement actually can lead to sparse networks and thereby to an increase in criminal activity. When criminal activity requires a certain degree of specialization, criminals will form sparse networks, which generate the highest level of crime and are the hardest to disrupt. I also show that heavy surveillance and large fines do not deter crime for these networks. The second essay examines the impact that residential location decisions have on economic outcomes of immigrants. About two thirds of the immigrants that arrived to the United States between 1997 and 2006 settled in six States only. Using a simultaneous-move game on residential choices I show that when all immigrants are unskilled they cluster in an enclave and earn very low wages, although they would be better off assimilating. Hence the enclave is `trap'. Introducing skill heterogeneity among immigrants reverses the result: the enclave equilibrium becomes socially preferred to assimilation.