Electronic Thesis or Dissertation

Computational tools grounded in algebraic topology, known collectively as topological data analysis (TDA), have been used for dimensionality-reduction to preserve salient and discriminating features in data. This faithful but compressed representation of data through TDA’s flagship method, persistent homology (PH), motivates its use to address the complexity, depth, and inefficiency issues present in privacy-preserving, homomorphic encryption (HE)-based machine learning (ML) models, which permit a data provider (often referred to as the Client) to outsource computational tasks on their encrypted data to a computationally-superior but semi-honest party (the Server). This work introduces efforts to adapt the well-established TDA-ML pipeline on encrypted data to realize the benefits TDA can provide to HE’s computational limitations as well as provide HE’s provable security on the sensitive data domains in which TDA has found success in (e.g., sequence, gene expression, imaging). The privacy-protecting technologies which could emerge from this foundational work will lead to direct improvements to the accessibility and equitability of health care systems. ML promises to reduce biases and improve accuracies of diagnoses, and enabling such models to act on sensitive biomedical data without exposing it will improve trustworthiness of these systems.

The purpose of this study is to understand how the relationship between generational cattle ranchers and conservation efforts is changing given rapid development and urbanization within the state of Florida over the past decade. To understand this relationship, the author used participant observation in conjunction with ethnographic interviews to work with ranchers and individuals from the Green Paths Foundation to draw a comprehensive picture of how development has affected ranching and conservation efforts in Florida. With the help of cultural consultants, I was able to make additional connections within the ranching and conservation communities that helped to provide further context to the question being investigated in this thesis and to understand how Florida’s Conservation Corridor mediates between ranchers and conservation imperatives.

In this study, the researcher uses quantitative methods to examine the extent to which Title I funding helps public schools with large populations of economically-disadvantaged students increase student academic achievement in reading and math in grades 3–5, and whether the leaders of these schools utilize specific social justice actions identified through a review of literature. The researcher gathers grade level student proficiency data for students in grades 3–5 and performs a bivariate correlation and a simple regression analysis to determine the extent to which identified schools receiving Title I funds are able to increase student proficiency rates over a three-year period. The researcher further analyzes the data through a social justice lens to determine possible social justice solutions to the persistent problem of the income-achievement gap.

Local social capital, defined as the level of community interaction and social participation of a region, has been theorized to positively affect economic outcomes and discourage opportunistic behaviors in various settings. I examine whether local social capital is related to positive outcomes for entrepreneurs and their financial backers in the settings of reward crowdfunding and small business lending.
In my first study, I look at how local social capital influences the creators of successful reward crowdfunding campaigns. These creators, in turn, may influence the sentiment of their investors, or backers, towards their projects through missed delivery deadlines and poor communication. With comments collected from successful Kickstarter crowdfunding pages, I use textual analysis to construct a measure of the sentiment of project backers following the fundraising deadline.

Rapid response and efficient damage assessment are life-or-death matters in the wake of natural disasters such as hurricanes and earthquakes. These events wreak havoc on infrastructure and properties and, most critically, endanger human lives. The timely and effective allocation of resources during such crises is imperative, necessitating meticulous planning based on the extent of damage incurred. This research presents an approach to automating the damage assessment using pre/post-disaster aerial images and computer vision. Recent advancements in disaster response strategies have encouraged researchers to harness the power of satellite and aerial imagery to assess the aftermath. Usually, due to the different characteristics between training datasets and available datasets in times of disasters, retraining the model to improve detection accuracy has been the norm, even though it is time and resource intensive. Our method surpasses conventional solutions and requires no retraining or fine-tuning on disaster-specific data. An existing model was retrained and improved on a diverse building damage dataset and demonstrably generalizes to new disaster scenarios. Having achieved higher performances compared to state of the art models, we determines our models real world applicability by using Hurricane Ian as our potent study grounds.

The rapid growth of digital transactions and the increasing sophistication of fraudulent activities have necessitated the development of robust and efficient fraud detection techniques, particularly in the financial and healthcare sectors. This dissertation focuses on the use of novel data reduction techniques for addressing the unique challenges associated with detecting fraud in highly imbalanced Big Data, with a specific emphasis on credit card transactions and Medicare claims. The highly imbalanced nature of these datasets, where fraudulent instances constitute less than one percent of the data, poses significant challenges for traditional machine learning algorithms. This dissertation explores novel data reduction techniques tailored for fraud detection in highly imbalanced Big Data. The primary objectives include developing efficient data preprocessing and feature selection methods to reduce data dimensionality while preserving the most informative features, investigating various machine learning algorithms for their effectiveness in handling imbalanced data, and evaluating the proposed techniques on real-world credit card and Medicare fraud datasets.
This dissertation covers a comprehensive examination of datasets, learners, experimental methodology, sampling techniques, feature selection techniques, and hybrid techniques. Key contributions include the analysis of performance metrics in the context of newly available Big Medicare Data, experiments using Big Medicare data, application of a novel ensemble supervised feature selection technique, and the combined application of data sampling and feature selection. The research demonstrates that, across both domains, the combined application of random undersampling and ensemble feature selection significantly improves classification performance.

Media literacy education has yet to adequately address the needs of parents as media literacy learners. Yet, a lack of understanding about media literacy’s meaning, for most people outside of the community of scholars and advocates engaged in its discourse, renders no clear pathway toward rectifying the omission. This study situates media, rather than media literacy, as the introductory topic, while seeking a conceptual gateway for parents as media literacy learners through insights gathered from mothers of adolescents on matters of quality, role, and impact within their home environments. Utilizing Bronfenbrenner’s bioecological theory of human development as a conceptual foundation, media use was framed here as a proximal process. Through exploratory mixed methods research design, data were collected from a quantitative questionnaire (n=363) and qualitative, semi-structured interviews (n=13) about three proximal process variations: (1) mother and adolescent engaged in joint media use, (2) the mother’s solitary media use, and (3) media use of their adolescent child(ren) when alone.
Quantitative results show that mothers evaluate the quality of media use in their home environments as more favorable than not for each proximal process. However, there was significant disagreement among the mothers regarding media use of their adolescent(s), when alone. Disagreement on this item guided identification of participants for the qualitative interviews.

This thesis examines the skeletal remains of two disabled adults collected from the Bogoz archaeological site (1100-1700) in Mugeni, Romania. Mugeni (in Hungarian, Bogoz) is home to an ethnic culture known as the Szekely, whose history has been lost (Bethard 2019, p. 254). This thesis conducts a microhistorical bioarchaeology of caregiving behaviors for Burial 13 and Burial 150 to concurrently reinsert disabled individuals into the historical narrative and to contribute to Szekely history.
Four theoretical backgrounds- microhistory, social bioarchaeology, osteobiography, and the Bioarchaeology of Care- are synthesized to organize analysis. First, this thesis documents biological identifiers, pathologies, mortuary treatment, and the physical, socio-cultural, and economic lifeways (Tilley & Schrenk 2017, p. 2). Then, models of care are developed to analyze multiscalar intersectionalities to understand the broader implications of medieval and early modern Transylvania (Peltonen 2001, p. 348; Walton 2008, p. 6). This approach will serve as an example for the continued investigations of care provisions for disabled and/or impaired persons, contributing to the historical narrative (Bethard et al. 2019, p. 267; Hosek 2019, p. 47).

This project examines some of the ethical consequences of posthumous publication of authors' unfinished works, private correspondence, and other materials, with the central example being the extensive catalogue of J.R.R. Tolkien published in the decades after this death by his son Christopher Tolkien. It builds a moral and philosophical framework for understanding the "posthumous harm" that can impact the deceased, for example when the desires they expressed in life are frustrated, or their reputations suffer damage when draft or private materials become public, especially for a wide audience. In the case of J.R.R. Tolkien, his Beowulf translation shows how an author's intentions for a work may actually be to not publish, as doing so contradicts their beliefs and values. Both literary executors and the consumer public that creates a market for such "new works" should more carefully evaluate the posthumous harm that posthumous publication can bring.

Herein, we discuss a novel method for the synthesis of decorated 2,5-dihydrofurans. The base promoted 5-endo-dig cyclization of non-Conia-ene propargyl ethers produces 2,2 disubstituted dihydrofurans. Central to the reaction is the presence of an acidic C–H bond which is activated by an adjacent aromatic heterocycle. The transformation is viable with a wide range of substituents, including N, O, and S containing heterocycles, substituted phenyl rings, and alkyl groups. The cyclization proceeds within 30 seconds at room temperature under the action of potassium tert-butoxide. This work stands apart from the current literature due to the absence of transition metal catalysts and/or harsh reaction conditions. A thorough mechanistic investigation is undertaken to better understand the nature of this unprecedented reaction.