University of Central Florida
University of Central Florida (UCF): STARS (Showcase of Text, Archives, Research & Scholarship)Not a member yet
166656 research outputs found
Sort by
Partisan Divides in Environmental Spending Attitudes: A Two-Level Hierarchical Analysis, 1973-2022
No full textPublic attitudes toward environmental spending have become increasingly divided along party lines, with sharp shifts over the past five decades. This thesis updates and expands on Johnson and Schwadel’s 2019 study by applying a two-level hierarchical linear model to General Social Survey data updated to include data from 2015-2022, capturing how political affiliation, education, race, and economic context interact with broader political and economic contexts to shape environmental attitudes over time. The results show that political affiliation remains the strongest and most reactive predictor of environmental spending attitudes. Republican respondents are significantly more likely to oppose environmental spending, especially under Democratic presidencies. In contrast, Democratic support remains relatively stable across changing political contexts. Education and Age also emerged as significant factors: individuals with more education and younger respondents are more likely to support environmental spending. Race had mixed statistical significance. Income, sex, and unemployment are not significant predictors on their own, but interactions indicate that attitudes may shift based on context. This study reinforces the central role of partisan identity in shaping environmental views and demonstrates how political and economic context can activate or dampen that divide. By incorporating new interaction terms and updated data, the model offers a clearer understanding of how these dynamics have evolved over time. The findings underscore the challenges of building bipartisan support for environmental policy and point to the importance of messaging that considers political context and reactive partisan patterns
High-Throughput Drug Screening For Disease-Modifying Arthritis Therapy
No full textCurrent therapy for arthritis predominantly focuses on the symptoms rather than the underlying causes of disease progression. Thus, it is necessary to identify novel drugs which target the underlying molecular basis of such diseases. Using high-throughput cell-based drug screening, we screened 458 different compounds, narrowed down from 811 starting compounds from the NIH Mechanistic Set. We selected drugs from the Mechanistic Set following Lipinski\u27s Rule of 5 to ensure optimal pharmacokinetic properties. The study used modified primary human chondrocytes (IIAM-PRG4Luc), which secrete a lubricin promoter-driven luciferase reporter. We use this luciferase expression as a proxy to measure lubricin, a compound that reduces friction between articular cartilage. These cells were grown and tested in a physioxic (5% oxygen) condition which has been shown to better model joint conditions within the body as opposed to typical tissue culture in atmospheric oxygen (20%). Cells formed cartilage aggregates (5000 cells) in 384 well plates with positive and negative controls. The drugs were diluted in cell media to 2µM and fed to cells 10 times over 21 days (3 weeks). The cells were imaged two times a week to track chondrogenesis. This assay was repeated in 96 well plates with the 8 hits we found in the 384 well assay. The aggregates were then tested for lubricin, fixed, and sectioned for histology and staining. We hope to identify a compound that may improve cartilage\u27s structural and tensile features and be a potential disease-modifying therapeutic for cartilagerelated diseases
Characterization of Electrical and Optical Properties of Europium Oxide
No full textOxide semiconductors with high optical transparency hold significant promise for a broad range of optical applications. Among them, europium oxide (Eu₂O₃ ) has garnered considerable attention for its potential in transistor devices, resistive random-access memory (RRAM), white light-emitting devices, and erasable optical storage systems. Additionally, Eu₂O₃ is employed in nanoscale applications, such as photoactive coatings and high-k dielectrics. Its advantageous dielectric properties and high transmittance make it particularly well-suited for capacitor fabrication. The material also exhibits resistive switching behavior and notable optical absorption, further enhancing its relevance for optoelectronic devices.
In this study, Eu₂O₃ thin films were deposited using RF magnetron sputtering at a fixed power of 50 W under varying pressure conditions at room temperature. The films were subsequently annealed in oxygen and nitrogen at 100 – 200 °C. The films annealed in Nitrogen and Oxygen showed an increase in dielectric constant. For optical studies, films are deposited on a quartz substrate. The bandgap of the film is extracted using transmission studies in the UV-visible range. X-ray diffraction was used to investigate Europium Oxide crystallization behavior by annealing samples with oxygen and nitrogen at 600 and 750 °C temperatures
Interpretation and Control of AI Model Behavior Through Direct Adjustment of Latent Representations
No full textThis dissertation investigates the structures and mechanisms underpinning the latent space representations that emerge within Generative Pretrained Transformer (GPT) models. Addressing the broader goal of enhancing AI trustworthiness through transparency, accountability, and controllability, we focus on techniques to understand, quantify, and manipulate these latent space representations. Through a series of analyses, we examine several chess-playing GPT models as controlled testbeds, leveraging their structured decision space to explore emergent representations and decision-making processes.
Key contributions include a mechanistic analysis of the attention heads and latent representations, the development of novel metrics for evaluating intervention outcomes, and the application of linear probe classifiers to decode and edit the model\u27s internal world representations. Analysis of the probe weight vectors reveals that the chess-playing GPT developed an emergent world model of the game that includes pieces, positions, and movement rules, and provides empirical support for the linear representation hypothesis—the idea that abstract concepts are encoded as specific directions in the model\u27s hidden state space. Complementary analysis of the hidden state vectors demonstrates that the model\u27s internal representations honor the Markovian property of chess.
Experimental results demonstrate that linear interventions can causally steer GPT outputs while preserving their semantic validity. Drawing on the dose-response analogy from medicine, we vary both the strength and position of interventions, showing that output quality is maximized when intervention strength follows an exponentially decaying schedule across token positions. Similar experiments using sparse autoencoders in place of linear probes yielded significantly poorer performance. These results highlight the effectiveness of simple linear probes as valuable tools for interpretability and control
Accurate and Efficient Orbit Probability Approximation Framework for Space Situational Awareness
No full textUncertainty Propagation in astrodynamics has gained importance in space situational awareness (SSA) problems such as space debris tracking, collision avoidance, and Cislunar operations. In this dissertation, the technique of Orbit Probability Approximation (OPA) is developed. OPA propagates orbital uncertainty using Liouville’s theorem with different functional approximations. First, OPA is formulated with Chebyshev polynomials to propagate the uncertainty on a geocentric planar orbit problem and then validated using two sources of satellite data: GRACE navigation data from the Jet Propulsion Laboratory (JPL) database, and FireOPAL ground-based observer provided by Lockheed Martin. In this validation process, OPA propagates uncertainty without using any measurements. Results indicate successful validation using GRACE navigation data for a low Earth orbit (LEO), and FireOPAL sensor tracking data for Yamal 202 in geosynchronous Earth orbit (GEO) and a rocket body of the Block-DM satellite in a highly elliptical orbit (HEO). Next, the technique of utilizing Radial Basis Functions (RBF) for uncertainty propagation is formulated and demonstrated on the geocentric planar orbit problem as well as a short-period L4 orbit in Cislunar space. Since RBFs allow the possibility to employ scattered nodes and dimension-wise shape parameters, the optimization of shape parameters and adaptive sampling strategies are explored to mitigate the curse of dimensionality. With cumulative integral as the chief error metric, OPA results show good agreement with Monte-Carlo and polynomial chaos expansion simulations. For example, in GRACE validation, OPA achieves 0.11% error from the true distribution. For the Cislunar orbit problem, OPA uses around 4300 nodes to achieve the same cumulative integral value as that of a Monte-Carlo simulation with 1 million points, a significant reduction in the computational cost. Future work explores the integration of OPA RBF approximation with Mamba, a state-space formulation-based recurrent neural network, to predict the future PDFs of nonlinear dynamical systems
\u27How to Mother\u27: Televisual Reimaginings of Black Matriarchy in the Real Housewives of Potomac
No full textSince the initial launch of The Real Housewives of Orange County in 2006, Bravo’s The Real Housewives franchise has soared to previously unimaginable heights. With 11 American installments, 30 international adaptations, and 25 current spin-offs, it is clear that even viewers who are relatively unfamiliar with The Real Housewives franchise may find themselves inadvertently swept up in their world due to the popularization of their catchphrases and gifs on social media platforms like Instagram, Twitter, and Facebook. In addition to carefully shaping popular culture, members of this franchise have even altered the conditions of the entertainment industry’s practices for the foreseeable future. Although The Real Housewives of Atlanta was the first Housewives show to feature a predominantly Black cast, The Real Housewives of Potomac found a new way to make their mark on Housewives history. In this thesis, I argue that a textual analysis of the sociocultural events found within The Real Housewives of Potomac reflect American society’s dissonant relationship with motherhood and by extension, the fragmented understanding of the Black female body. Through this connection, I challenge the synonymous bond between isolation and housewifery, positing that the community formed by castmates’ interpersonal connections is not only imperative to their success in the series but also as Black women
Advancing Anomaly Detection with Robust and Graph-Based Learning Methods: From Support Vector Data Description to Graph Neural Networks.
No full textAnomaly detection is crucial across various domains, particularly in handling highly skewed datasets where only normal operating conditions are available for training. To effectively identify abnormal events, specialized one-class classifiers have been developed. This dissertation explores robust and scalable anomaly detection methods, focusing on enhancing support vector techniques to accommodate complex data structures like graphs. The first study introduces the Robust Support Vector Data Description (RSVDD) model, which improves standard SVDD by incorporating a rescaled hinge loss function, making it more resistant to outliers. Using a half-quadratic optimization method, RSVDD dynamically adjusts the influence of each data point, leading to improved anomaly detection in both synthetic and real-world datasets. The second study addresses anomaly detection in graph-structured data, where node dependencies introduce additional challenges. The proposed Least Squares One-Class Graph Neural Network (LS-OCGNN) integrates Graph Neural Networks (GNNs) with a least-squares hypersphere learning approach. This novel framework enhances efficiency by leveraging a closed-form solution for hypersphere radius estimation, reducing Type II errors and improving sensitivity to complex anomalies, as demonstrated on datasets such as the Cora citation network. The final study extends one-class classification by introducing Least Squares Support Vector Machine with Graph Neural Networks (LSOCSVM-GNN), a scalable framework for detecting anomalies in attributed networks. By integrating node features and graph structure, LSOCSVM-GNN overcomes limitations of traditional SVM-based methods in graph-based scenarios. Together, these studies advance one-class classification by combining classical support vector approaches with modern graph-based learning frameworks. The proposed methods offer scalable, high-performance anomaly detection solutions with practical applications in cybersecurity, fraud detection, and social network analysis
Emojis as Text Technology: Between Visual Paradigm and Cultural Revolution
No full textEmojis, as a broader communication phenomenon, have gained widespread usage, significant adaptation, and considerable academic attention due to their role in providing semiotic cues and online gestures within computer-mediated communication. This study aims to trace their development and evolution using the text technologies framework. This approach examines the systematic use of emojis, with particular emphasis on their extensive evolution by historically contextualizing their usage, trends, and cultural interactions. Anchored in this conceptual approach, the study draws on the taxonomy of text technology developed by Treharne and Willan, analyzing the evolution of emojis through four key dimensions: intentionality, materiality, functionality, and cultural value. This historical investigation explores the development, technical standardization, agency and governance, and participatory cultural dynamics and practices associated with emojis. Guided by two central lines of inquiry, the study presents an integrated perspective that moves beyond the visual and semantic functions of emoji to reveal the structural forces and sociotechnical systems that shape their form, meaning, and circulation within digital media. The first line of inquiry establishes a foundation for exploring the transformative impact of emojis, shedding light on their role in shaping self-expression, fostering social connections, and influencing users’ navigational behavior in digital environments. Building upon this, the second line of inquiry emphasizes the technological dimensions of emoji communication, examining how technical affordances and constrain along with design parameters actively shape and redefine interpretation and interaction across diverse computer-mediated communication platforms. As a result of tracing the historical trajectory of emojis within the Unicode framework, this study explores the interplay between technological innovation and evolving communication practices, offering a comprehensive perspective on emojis as dynamic tool situated at the intersection of language, culture, and technology
The Hilbert Series of Paths, Cycles, and Related Graphs
No full textThe Hilbert Series of a finitely-generated graded R-module, M, is a series which is often given in the form a rational function in the variable, t. This series encodes a great many invariant properties of the module M. In this dissertation, I study the Hilbert series and related invariants of the graph rings for paths and cycles. By utilizing a result of Kyle Trainor, I am able to examine the Hilbert series and the related invariants of these graph rings recursively through second and higher-order difference equations. This technique allows me to extract information about the Hilbert series that has not been previously well known, and for which other techniques are limited
From Failure To Fidelity: Enabling Scalable Sim2real Lidar Perception Through Realistic Digital Twins
No full textLiDAR technology plays a key role in enabling intelligent perception across urban systems. Its ability to capture accurate 3D data regardless of lighting conditions makes it well-suited for applications in autonomous vehicles, traffic monitoring, security, and privacy-aware smart city infrastructure, offering advantages over camera-based systems in depth, precision, and reliability. However, scaling deployment faces substantial challenges due to the time, effort, and cost of gathering and labeling LiDAR data used to train and test perception algorithms. This dissertation addresses these challenges by tracing a comprehensive research trajectory from early failures to a successful, fidelity-driven solution. Initially, (1) it explores transfer learning, highlighting its limitations in generalization, and (2) develops a self-supervised method leveraging teacher-student modeling to train deep neural object detectors. Despite promising results in constrained settings, these approaches showed limited scalability and inconsistent real-world performance. In response, the dissertation studies Sim2Real learning, (3) investigating limitations, and (4) presents a novel scalable Sim2Real learning framework. This framework uses high-fidelity (HiFi) digital twins (DTs) that replicate real locations with publicly available geospatial data, preserving structural and contextual details such as background geometry, road characteristics, and traffic distributions. By simulating sensors in these HiFi DTs, in-domain LiDAR datasets are synthesized. These synthetic datasets can train perception models that consistently match or exceed the performance of models trained on real data. In addition to these methodological advances, the dissertation contributes to research through (i) LiGuard: streamlined open-source LiDAR data processing and visualization software for rapid research, (ii) UrbanTwin: open-source digital-twins of real locations advancing research beyond perception, and (iii) open-source synthetic LiDAR datasets. By systematically documenting and mitigating foundational failures, this work provides a blueprint for building robust, scalable LiDAR perception systems for real-world ITS deployment