Digital Commons Network^™

Machine Learning Techniques For Topic Detection And Authorship Attribution In Textual Data, Fereshteh Jafariakinabad

Electronic Theses and Dissertations, 2020-

The unprecedented expansion of user-generated content in recent years demands more attempts of information filtering in order to extract high-quality information from the huge amount of available data. In this dissertation, we begin with a focus on topic detection from microblog streams, which is the first step toward monitoring and summarizing social data. Then we shift our focus to the authorship attribution task, which is a sub-area of computational stylometry. It is worth mentioning that determining the style of a document is orthogonal to determining its topic, since the document features which capture the style are mainly independent of its …

Studying Users Interactions And Behavior In Social Media Using Natural Language Processing, Sultan Alshamrani

Electronic Theses and Dissertations, 2020-

Social media platforms have been growing at a rapid pace, attracting users' engagement with the online content due to their convenience facilitated by many useful features. Such platforms provide users with interactive options such as likes, dislikes as well as a way of expressing their opinions in the form of text (i.e., comments). As more people engage in different social media platforms, such platforms will increase in both size and importance. This growth in social media data is becoming a vital new area for scholars and researchers to explore this new form of communication. The huge data from social media …

Directional Spectral Solar Energy For Building Performance: From Simulation To Cyber-Physical Prototype, Joseph Del Rocco

Electronic Theses and Dissertations, 2020-

The original research and development in this dissertation contributes to the field of building performance by actively harnessing a wider spectrum of directional solar radiation for use in buildings. Solar radiation (energy) is often grouped by wavelength measurement into the spectra ultraviolet (UV), visible (light), and short and long-wave infrared (heat) on the electromagnetic spectrum. While some of this energy is directly absorbed or deflected by our atmosphere, most of it passes through, scatters about, and collides with our planet. Modern building performance simulations, tools, and control systems often oversimplify this energy into scalar values for light and heat, when …

Efficient Data Structures For Text Processing Applications, Paniz Abedin

Electronic Theses and Dissertations, 2020-

This thesis is devoted to designing and analyzing efficient text indexing data structures and associated algorithms for processing text data. The general problem is to preprocess a given text or a collection of texts into a space-efficient index to quickly answer various queries on this data. Basic queries such as counting/reporting a given pattern's occurrences as substrings of the original text are useful in modeling critical bioinformatics applications. This line of research has witnessed many breakthroughs, such as the suffix trees, suffix arrays, FM-index, etc. In this work, we revisit the following problems: 1. The Heaviest Induced Ancestors problem 2. …

Quantifiability: Concurrent Correctness From First Principles, Victor Cook

Electronic Theses and Dissertations, 2020-

Architectural imperatives due to the slowing of Moore's Law, the broad acceptance of relaxed semantics and the O(n!) worst case verification complexity of sequential histories motivate a new approach to concurrent correctness. Desiderata for a new correctness condition are that it be independent of sequential histories, compositional over objects, flexible as to timing, modular as to semantics and free of inherent locking or waiting. This dissertation proposes Quantifiability, a novel correctness condition based on intuitive first principles. Quantifiablity is formally defined with its system model. Useful properties of quantifiability such as compositionality, measurablility and observational refinement are demonstrated. Quantifiability models …

Capsule Networks For Video Understanding, Kevin Duarte

Electronic Theses and Dissertations, 2020-

With the increase of videos available online, it is more important than ever to learn how to process and understand video data. Although convolutional neural networks have revolutionized the representation learning from images and videos, they do not explicitly model entities within the given input. It would be useful for learned models to be able to represent part-to-whole relationships within a given image or video. To this end, a novel neural network architecture - capsule networks - has been proposed. Capsule networks add extra structure to allow for the modeling of entities and has shown great promise when applied to …

The Social And Behavioral Influences Of Interactions With Virtual Dogs As Embodied Agents In Augmented And Virtual Reality, Nahal Norouzi

Electronic Theses and Dissertations, 2020-

Intelligent virtual agents (IVAs) have been researched for years and recently many of these IVAs have become commercialized and widely used by many individuals as intelligent personal assistants. The majority of these IVAs are anthropomorphic, and many are developed to resemble real humans entirely. However, real humans do not interact only with other humans in the real world, and many benefit from interactions with non-human entities. A prime example is human interactions with animals, such as dogs. Humans and dogs share a historical bond that goes back thousands of years. In the past 30 years, there has been a great …

Human Behavior In Domestic Environments: Prediction And Applications, Sharare Zehtabian

Electronic Theses and Dissertations, 2020-

A longstanding goal of human behavior science is to model and predict how humans interact with each other or with other systems. Such models are beneficial and have many applications, including designing and implementing assistive technologies, improving users' experiences and quality of life and making better decisions to create public policies. Behavior is highly complex due to uncertainties and a lack of scientific tools to measure it. Hence prediction of human behavior cannot be 100% accurate. However, prediction is also not hopeless because the biological needs, as well as cultural conventions (for instance, regarding meal times) set the general patterns …

Self-Governed Iot Networks: A Blockchain-Based Framework, Mehrdad Salimitari

Electronic Theses and Dissertations, 2020-

In the modern-day, IoT-enabled devices and sensors are growingly deployed in a variety of applications in consumer, commercial, industrial, and infrastructure spaces. Thus, decentralization of such networks for reducing the maintenance and repair costs is significantly gaining attention. In this dissertation, we investigate a variety of mathematical tools and technological advancements to design a secure, fully decentralized, and self-governed IoT network. On this path, the main challenge is ensuring the data integrity in IoT networks, securing the devices against a variety of attack scenarios, and preserving the privacy of the users. We begin by proposing a prospect theoretic method for …

A Deep Learning Approach For Learning Human Gait Signature, Alexander Matasa

Electronic Theses and Dissertations, 2020-

With advancements in biometric securities, focus has increased on utilizing gait as a means of recognition. Gait describes the unique walking pattern present in humans and has shown promising results in person re-identification tasks. Unlike other biometric features, gait is unique in that it is a subconscious behavior minimizing the risk of purposeful obfuscation. In this research, we first cover supervised approaches showing that current methods fail to learn a unique signature that describes the motion of a subject. Rather they extract frame-based feature information which is then aggregated. While these methods have shown to be effective, they do not …

Secure And Trustworthy Hardware And Machine Learning Systems For Internet Of Things, Shayan Taheri

Electronic Theses and Dissertations, 2020-

The advancements on the Internet have enabled connecting more devices into this technology every day. This great connectivity has led to the introduction of the internet of things (IoTs) that is a great bed for engagement of all new technologies for computing devices and systems. Nowadays, the IoT devices and systems have applications in many sensitive areas including military systems. These challenges target hardware and software elements of IoT devices and systems. Integration of hardware and software elements leads to hardware systems and software systems in the IoT platforms, respectively. A recent trend for the hardware systems is making them …

Spatial-Temporal Representation Learning: Concepts, Algorithms And Applications, Pengyang Wang

Electronic Theses and Dissertations, 2020-

Recent years have witnessed the flourish of Internet-of-Things (IoT), in which sensors connect spatial entities to constitute complex Cyber-Physical Systems (CPSs). In this setting, spatial-temporal data becomes increasingly available. Mining spatial-temporal data can reveal holistic user and system structures, dynamics, and semantics of the underlying CPSs, including identifying trends, forecasting future behavior, and detecting anomalies. However, obtaining effective representations over spatial-temporal data remains a big challenge for the following reasons: (1) on the one hand, traditional manual feature design is labor-intensive and time-consuming facing the complex and huge volumes of spatial-temporal data; (2) on the other hand, as an emerging …

Visual Learning Beyond Human Curated Datasets, Muhammad Abdullah Jamal

Electronic Theses and Dissertations, 2020-

The success of deep neural networks in a variety of computer vision tasks heavily relies on large- scale datasets. However, it is expensive to manually acquire labels for large datasets. Given the human annotation cost and scarcity of data, the challenge is to learn efficiently with insufficiently labeled data. In this dissertation, we propose several approaches towards data-efficient learning in the context of few-shot learning, long-tailed visual recognition, and unsupervised and semi-supervised learning. In the first part, we propose a novel paradigm of Task-Agnostic Meta- Learning (TAML) algorithms to improve few-shot learning. Furthermore, in the second part, we analyze the …

Unsupervised Meta-Learning, Siavash Khodadadeh

Electronic Theses and Dissertations, 2020-

Deep learning has achieved classification performance matching or exceeding the human one, as long as plentiful labeled training samples are available. However, the performance on few-shot learning, where the classifier had seen only several or possibly only one sample of the class is still significantly below human performance. Recently, a type of algorithm called meta-learning achieved impressive performance for few-shot learning. However, meta-learning requires a large dataset of labeled tasks closely related to the test task. The work described in this dissertation outlines techniques that significantly reduce the need for expensive and scarce labeled data in the meta-learning phase. Our …

Evaluating Augmented Reality Tools For Physics Education, Corey Pittman

Electronic Theses and Dissertations, 2020-

While we are in the midst of a renaissance of interest in augmented reality (AR), there remain a small number of application domains that have seen significant development. One domain that often benefits from additional visualization capabilities is education, specifically physics and other sciences. This paper summarizes interviews with secondary school educators about their experience with AR and their most desired features. Three prototypes were created which were used to collect usability information from students and educators about their preferences for AR applications in their physics courses. Additionally, we introduce the concept of Environmental Integration, a novel method of defining …

Contextual Understanding Of Sequential Data Across Multiple Modalities, Sangwoo Cho

Electronic Theses and Dissertations, 2020-

In recent years, progress in computing and networking has made it possible to collect large volumes of data for various different applications in data mining and data analytics using machine learning methods. Data may come from different sources and in different shapes and forms depending on their inherent nature and the acquisition process. In this dissertation, we focus specifically on sequential data, which have been exponentially growing in recent years on platforms such as YouTube, social media, news agency sites, and other platforms. An important characteristic of sequential data is the inherent causal structure with latent patterns that can be …

Analyzing The Blockchain Attack Surface: A Top-Down Approach, Muhammad Saad

Electronic Theses and Dissertations, 2020-

Blockchains enable secure asset exchange in a distributed system, thereby facilitating innovative applications such as cryptocurrencies and smart contracts. Although the cryptographic constructs of blockchains are highly secure, however, their practical deployments are vulnerable to various attacks due to their application-specific policies, and their peer-to-peer (P2P) network intricacies. In this work, we take a top-down approach towards exploring those attacks, starting with the application-specific abuse of blockchain-based cryptocurrencies and concluding with the network conditions that violate the blockchain consistency. In the top-down approach, we first analyze the application-specific abuse of blockchain-based cryptocurrencies by uncovering (1) covert cryptocurrency mining in the …

Family Communication: Examining The Differing Perceptions Of Parents And Teens Regarding Online Safety Communication, Tara Rutkowski

Honors Undergraduate Theses

The opportunity for online engagement increases possible exposure to potentially risky behaviors for teens, which may have significant negative consequences (Hair et al., 2009). Effective family communication about online safety can help reduce the risky adolescent behavior and limit the consequences after it occurs. This paper contributes a theory of communication factors that positively influence teen and parent perception of communication about online safety and provides design implications based on those findings. Previous work identified gaps in family communication, however, this study seeks to empirically identify factors that would close the communication gap from the perspective of both teens and …

Reviving Mozart With Intelligence Duplication, Jacob E. Galajda

Honors Undergraduate Theses

Deep learning has been applied to many problems that are too complex to solve through an algorithm. Most of these problems have not required the specific expertise of a certain individual or group; most applied networks learn information that is shared across humans intuitively. Deep learning has encountered very few problems that would require the expertise of a certain individual or group to solve, and there has yet to be a defined class of networks capable of achieving this. Such networks could duplicate the intelligence of a person relative to a specific task, such as their writing style or music …

Binary State Distance Vector Routing: A Protocol For Near-Unicast Forwarding In Partitioned Networks, Ammar Farooq

Electronic Theses and Dissertations, 2020-

Ad-hoc networks are highly dynamic and can be disconnected/partitioned during their operations, particularly in delay-tolerant networks (DTNs) where infrastructure support is not entirely available. Even after several decades of research on DTN routing, there is still a need for routing protocols that operate effectively in network environments where disconnections, delays, and resource scarcity are common. Traditionally, DTN routing protocols use an epidemic routing strategy, where multiple copies of packets get forwarded to increase network reachability. However, these flooding-based strategies are seldom suitable in resource-constrained network settings. Other prominent DTN routing designs use distance vectors (DVs) that summarize global network reachability …

Exploring Relationships Between Ground And Aerial Views By Synthesis And Matching, Krishna Regmi

Electronic Theses and Dissertations, 2020-

Cross-view images, referring to the images taken from aerial and street views, contain drastically differing representations of the same scene of a given location. Due to the differences in the camera viewpoints of ground and aerial images the same semantic concepts in the two viewpoints look very different. Therefore the problem of relating them is very challenging. Thus, it becomes crucial to explore the cross-view relations and learn appropriate representations such that images from these two domains can be associated. In this dissertation we explore the relationship between ground and aerial views by synthesis and matching. First, we explore supervised …

Fpga-Augmented Secure Crash-Consistent Non-Volatile Memory, Yu Zou

Electronic Theses and Dissertations, 2020-

Emerging byte-addressable Non-Volatile Memory (NVM) technology, although promising superior memory density and ultra-low energy consumption, poses unique challenges to achieving persistent data privacy and computing security, both of which are critically important to the embedded and IoT applications. Specifically, to successfully restore NVMs to their working states after unexpected system crashes or power failure, maintaining and recovering all the necessary security-related metadata can severely increase memory traffic, degrade runtime performance, exacerbate write endurance problem, and demand costly hardware changes to off-the-shelf processors. In this thesis, we summarize and expand upon two of our innovative works, ARES and HERMES, to design …

Towards Enabling Explanation In Safety-Critical Artificial Intelligence Systems, Andy Michel

Electronic Theses and Dissertations, 2020-

With the advancement of accelerated hardware in recent years, there has been a surge in the development and application of intelligent systems. Deep learning systems, in particular, have shown exciting results in a wide range of tasks: classification, detection, and recognition. Despite these remarkable achievements, there remains an active research area that aims to increase the robustness of those systems in critical domains. Deep learning algorithms have proven to be brittle against adversarial attacks. That is, carefully crafted adversarial inputs can consistently trigger an erroneous prediction from a network model. Hence the motivation of this dissertation, we study prominent adversarial …

Towards Improving The Robustness Of Neural Abstractive Summarization, Kaiqiang Song

Electronic Theses and Dissertations, 2020-

Recent deep learning and sequence-to-sequence learning technology have produced impressive results on automatic summarization. However, the models have limited insights on the underlying language and it remains challenging for system-generated summaries to be truthful to the original input or cover the most important information. This is especially the case for generating abstractive summaries using neural models. My work aims for a flexible and controllable summarization system that can be adapted to cater to different scenarios. It is designed to incorporate linguistic structure information into deep neural networks, have the capability to produce abstracts by re-using a varying amount of source …

Spatio-Temporal Representation For Reasoning With Action Genome, Kesar Tumkur Narasimhamurthy

Electronic Theses and Dissertations, 2020-

Representing Spatio-temporal information in videos has proven to be a difficult task compared to action recognition in videos involving multiple actions. A single activity consists many smaller actions that can provide a better understanding of the activity. This paper tries to represent the varying information in a scene-graph format in order to answer temporal questions to obtain improved insights for the video, resulting in a directed temporal information graph. This project will use the Action Genome dataset, which is a variation of the charades dataset, to capture pairwise relationships in a graph. The model performs significantly better than the benchmark …

Synchronization And Analysis Of Multimodal Medical Data, Nafisa N. Mostofa

Honors Undergraduate Theses

The United States suffers from a significant disparity in the availability of the medical resources and expertise among different regions of the country. Patients in rural areas may not have the opportunity to consult with a physician until their disease progresses to later stages, resulting in a considerable decrease in quality of life. Advances in telemedicine systems that can provide remote communication, medical data acquisition, and medical data analysis promise a significant improvement to early access to medical care and diagnoses for disadvantaged individuals.

In this thesis, we make several contributions on topics that contribute to the improvement of telemedicine …

Fine-Grained Lower Bounds For Problems On Strings And Graphs, Gary Thomas Hoppenworth

Honors Undergraduate Theses

The motivation of this thesis is to present new lower bounds for important computational problems on strings and graphs, conditioned on plausible conjectures in theoretical computer science. These lower bounds, called conditional lower bounds, are a topic of immense interest in the field of fine-grained complexity, which aims to develop a better understanding of the hardness of problems that can be solved in polynomial time. In this thesis, we give new conditional lower bounds for four interesting computational problems: the median and center string edit distance problems, the pattern matching on labeled graphs problem, and the subtree isomorphism problem. These …

Recapture: A Virtual Reality Interactive Narrative Experience Concerning Perspectives And Self-Reflection, Indira Avendano

Honors Undergraduate Theses

This project presents a virtual reality (VR) Interactive Narrative aiming to leave users reflecting on the perspectives one chooses to view life through. The narrative is driven by interactions designed using the concept of procedural rhetoric, which explores how rules and mechanics in games can persuade people about an idea, and Shin's cognitive model, which presents a dynamic view of immersion in VR. The persuasive nature of procedural rhetoric in combination with immersion techniques such as tangible interfaces and first-person elements of VR can effectively work together to immerse users into a compelling narrative experience with an intended emotional response …

Improving Matching And Classification Through Deep Learning Of Structure And Varying Illumination, Sarah Braeger

Electronic Theses and Dissertations, 2020-

Convolutional networks have driven major advances in computer vision in recent years. The design of deep architectures, loss functions, and the curation of large, diverse datasets have furthered progress in many applied computer vision tasks. How data is represented to a network guides feature discovery and must be carefully considered in order to maximize performance on any applied task. We introduce novel input representations and associated architectural techniques to better utilize them such as complementary loss terms and network structure. We demonstrate the impact of these approaches on classification and matching tasks which involve shape and varied illumination. We show …

Algorithms And Lower Bounds For Ordering Problems On Strings, Daniel Gibney

Electronic Theses and Dissertations, 2020-

This dissertation presents novel algorithms and conditional lower bounds for a collection of string and text-compression-related problems. These results are unified under the theme of ordering constraint satisfaction. Utilizing the connections to ordering constraint satisfaction, we provide hardness results and algorithms for the following: recognizing a type of labeled graph amenable to text-indexing known as Wheeler graphs, minimizing the number of maximal unary substrings occurring in the Burrows-Wheeler Transformation of a text, minimizing the number of factors occurring in the Lyndon factorization of a text, and finding an optimal reference string for relative Lempel-Ziv encoding.