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
4 results
Search Results
Item TOWARDS EFFICIENT PRESENTATION AND INTERACTION IN VISUAL DATA ANALYSIS(2019) Cui, Zhe; JaJa, Joseph; Elmqvist, Niklas; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The "data explosion'' since the era of the Internet has increased data size tremendously, from several hundred Megabytes to millions of Terabytes. Large amounts of data may not fit into memory, and a proper way of handling and processing the data is necessary. Besides, analyses of such large scale data requires complex and time consuming algorithms. On the other hand, humans play an important role in steering and driving the data analysis, while there are often times when people have a hard time getting an overview of the data or knowing which analysis to run. Sometimes they may not even know where to start. There is a huge gap between the data and understanding. An intuitive way to facilitate data analysis is to visualize it. Visualization is understandable and illustrative, while using it to support fast and rapid data exploration of large scale datasets has been a challenge for a long time. In this dissertation, we aim to facilitate efficient visual data exploration of large scale datasets from two perspectives: efficiency and interaction. The former indicates how users could understand the data efficiently, this depends on various factors, such as how fast data is processed and how data is presented, while the latter focuses more on the users: how they deal with the data and why they interact with the system in a particular way. In order to improve the efficiency of data exploration, we have looked into two steps in the visualization pipeline: rendering and processing (computations). We first address visualization rendering of large dataset through a thorough evaluation of web-based visualization performance. We evaluate and understand the page loading effects of Scalable Vector Graphics (SVG), a popular image format for interactive visualization on the web browsers. To understand the scalability of individual elements in SVG based visualization, we conduct performance tests on different types of charts, in different phases of rendering process. From the results, we have figured out optimization techniques and guidelines to achieve better performance when rendering SVG visualization. Secondly, we present a pure browser based distributed computing framework (VisHive) that exploits computational power from co-located idle devices for visualization. The VisHive framework speeds up web-based visualization, which is originally designed for single computer and cannot make use of additional computational resources on the client side. It takes advantage of multiple devices that today's users often have access to. VisHive constructs visualization applications that can transparently connect multiple devices into an ad-hoc cluster for local computation. It requires no specific software to be downloaded for setup. To achieve a more interactive data analysis process, we first propose a proactive visual analytics system (DataSite) that enable users to analyze the data smoothly with a list of pre-defined algorithms. DataSite provides results through selecting and executing computations using automatic server-side computation. It utilizes computational resources exhaustively during data analysis to reduce the burden of human thinking. Analyzing results identified by these background processes are surfaced as status updates in a feed on the front-end, akin to posts in a social media feed. DataSite effectively turns data analysis into a conversation between the user and the computer, thereby reducing the cognitive load and domain knowledge requirements on users. Next we apply the concept of proactive data analysis to genomic data, and explore how to improve data analysis through adaptive computations in bioinformatics domain. We build Epiviz Feed, a web application that supports proactive visual and statistical analysis of genomic data. It addresses common and popular biological questions that may be asked by the analyst, and shortens the time of processing and analyzing the data with automatic computations. We further present a computational steering mechanism for visual analytics that prioritizes computations performed on the dataset leveraging the analyst's navigational behavior in the data. The web-based system, called Sherpa, provides computational modules for genomic data analysis, where independent algorithms calculate test statistics relevant to biological inferences about gene regulation in various tumor types and their corresponding normal tissues.Item SMART STRUCTURAL CONDITION ASSESSMENT METHODS FOR CIVIL INFRASTRUCTURES USING DEEP LEARNING ALGORITHM(2018) Liu, Heng; Zhang, Yunfeng; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Smart Structural Health Monitoring (SHM) technique capable of automated and accurate structural health condition assessment is appealing since civil infrastructural resilience can be enhanced by reducing the uncertainty involved in the process of assessing the condition state of civil infrastructures and carrying out subsequent retrofit work. Over the last decade, deep learning has seen impressive success in traditional pattern recognition applications historically faced with long-time challenges, which motivates the current research in integrating the advancement of deep learning into SHM applications. This dissertation research aims to accomplish the overall goal of establishing a smart SHM technique based on deep learning algorithm, which will achieve automated structural health condition assessment and condition rating prediction for civil infrastructures. A literate review on structural health condition assessment technologies commonly used for civil infrastructures was first conducted to identify the special need of the proposed method. Deep learning algorithms were also reviewed, with a focus on pattern recognition application, especially in the computer vision field in which deep learning algorithms have reported great success in traditionally challenging tasks. Subsequently, a technical procedure is established to incorporate a particular type of deep learning algorithm, termed Convolutional Neural Network which was found behind the many success seen in computer vision applications, into smart SHM technologies. The proposed method was first demonstrated and validated on an SHM application problem that uses image data for structural steel condition assessment. Further study was performed on time series data including vibration data and guided Lamb wave signals for two types of SHM applications - brace damage detection in concentrically braced frame structures or nondestructive evaluation (NDE) of thin plate structures. Additionally, discrete data (neither images nor time series data), such as the bridge condition rating data from National Bridge Inventory (NBI) data repository, was also investigated for the application of bridge condition forecasting. The study results indicated that the proposed method is very promising as a data-driven structural health condition assessment technique for civil infrastructures, based on research findings in the four distinct SHM case studies in this dissertation.Item ENGINEERING DIGITAL SHARING PLATFORM TO CREATE SOCIAL CONTAGION: EVIDENCE FROM LARGE SCALE RANDOMIZED FIELD EXPERIMENTS(2016) Sun, Tianshu; Viswanathan, Siva; Business and Management: Decision & Information Technologies; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Peer-to-peer information sharing has fundamentally changed customer decision-making process. Recent developments in information technologies have enabled digital sharing platforms to influence various granular aspects of the information sharing process. Despite the growing importance of digital information sharing, little research has examined the optimal design choices for a platform seeking to maximize returns from information sharing. My dissertation seeks to fill this gap. Specifically, I study novel interventions that can be implemented by the platform at different stages of the information sharing. In collaboration with a leading for-profit platform and a non-profit platform, I conduct three large-scale field experiments to causally identify the impact of these interventions on customers’ sharing behaviors as well as the sharing outcomes. The first essay examines whether and how a firm can enhance social contagion by simply varying the message shared by customers with their friends. Using a large randomized field experiment, I find that i) adding only information about the sender’s purchase status increases the likelihood of recipients’ purchase; ii) adding only information about referral reward increases recipients’ follow-up referrals; and iii) adding information about both the sender’s purchase as well as the referral rewards increases neither the likelihood of purchase nor follow-up referrals. I then discuss the underlying mechanisms. The second essay studies whether and how a firm can design unconditional incentive to engage customers who already reveal willingness to share. I conduct a field experiment to examine the impact of incentive design on sender’s purchase as well as further referral behavior. I find evidence that incentive structure has a significant, but interestingly opposing, impact on both outcomes. The results also provide insights about senders’ motives in sharing. The third essay examines whether and how a non-profit platform can use mobile messaging to leverage recipients’ social ties to encourage blood donation. I design a large field experiment to causally identify the impact of different types of information and incentives on donor’s self-donation and group donation behavior. My results show that non-profits can stimulate group effect and increase blood donation, but only with group reward. Such group reward works by motivating a different donor population. In summary, the findings from the three studies will offer valuable insights for platforms and social enterprises on how to engineer digital platforms to create social contagion. The rich data from randomized experiments and complementary sources (archive and survey) also allows me to test the underlying mechanism at work. In this way, my dissertation provides both managerial implication and theoretical contribution to the phenomenon of peer-to-peer information sharing.Item MINIMIZATION OF RESOURCE CONSUMPTION THROUGH WORKLOAD CONSOLIDATION IN LARGE-SCALE DISTRIBUTED DATA PLATFORMS(2014) Kayyoor, Ashwin Kumar; Deshpande, Amol; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The rapid increase in the data volumes encountered in many application domains has led to widespread adoption of parallel and distributed data management systems like parallel databases and MapReduce-based frameworks (e.g., Hadoop) in recent years. Use of such parallel and distributed frameworks is expected to accelerate in the coming years, putting further strain on already-scarce resources like compute power, network bandwidth, and energy. To reduce total execution times, there is a trend towards increasing execution parallelism by spreading out data across a large number of machines. However, this often increases the total resource consumption, and especially energy consumption, significantly because of process startup costs and other overheads (e.g., communication overheads). In this dissertation, we develop several data management techniques to minimize resource consumption through workload consolidation. In this dissertation, we introduce a key metric called query span, i.e., number of machines involved in the execution of a query or a job. In order to minimize the per query resource consumption we propose to minimize query span. To that end, we develop several workload-driven data partitioning and replica selection algorithms that attempt to minimize the average query span by exploiting the fact that most distributed environments need to use replication for fault tolerance. Extensive experiments on various datasets show that judicious data placement and replication can dramatically reduce the average query spans resulting in significant reductions in resource consumption. We show our results primarily on two applications, distributed data warehouse system and distributed information retrieval. In the first case, we show that minimizing average query spans can minimize overall resource consumption for a given workload and can also improve the performance of complex analytical queries. In the second case, our approach minimizes the overall search cost as well as effectively trades off search cost with load imbalance. The best case of resource efficiency for any underlying data processing system is achieved when the job or the query can be run efficiently on a single machine (i.e., query span=1). In the final part of dissertation, we discuss an in-memory MapReduce system optimized for performing complex analytics tasks on input data sizes that fit in a single machine's memory. We argue that systems like Hadoop that are designed to operate across a large number of machines are not optimal in performance for small and medium sized complex analytics tasks because of high startup costs, heavy disk activity, and wasteful checkpointing. We have developed a prototype runtime called HONE that is API compatible with standard (distributed) Hadoop. In other words, we can take existing Hadoop code and run it, without modification, on a multi-core shared memory machine. This allows us to take existing Hadoop algorithms and find the most suitable runtime environment for execution on datasets of varying sizes. Overall, in this dissertation, our key contributions in this work include identification of key metric query span and its relationship with overall resource consumption in scale-out architectures. We introduce several workload-aware techniques to optimize this key metric. We go on to demonstrate the effectiveness of query span minimization on different application scenarios. In order to take advantage of scale-up architectures effectively we develop novel in-memory MapReduce system HONE for single machine. Our thorough experiments on real and synthetic datasets demonstrate the efficacy of our proposed approaches.