Digital Commons @ Florida Atlantic
Not a member yet
400 research outputs found
Sort by
MOVIE GENRE CLASSIFICATION USING SCRIPT TEXTS
Get PDFGenres are used to classify movies so that they can be grouped with others that have similar themes and structures. These classifications are categories created by humans. In the process of creating a movie, a script is often the first creation to write and share ideas about a topic. The script contains large amounts of text that is used to describe the dialog, setting and direction of the film. Although the script contains important information for the film, the amount of text can present a challenge for machine learning algorithms. Often in studies on film classification, if text is used, the script is not chosen, but rather a review or a synopsis instead. The text in these is often smaller and easier to train on. However, a script presents some benefits as an input for model training. Being an earlier production in the process of creating a film, the script could be the only data available. The script could be available before the film is even created. In this study, we propose a method for using movie scripts to build models to classify a movie’s genre, as either a “Comedy” or a “Drama.” Our goal is to understand whether and to what extent movie scripts can be used to predict its genre. To better handle learning in this problem space, a model optimization process is proposed using data cleaning, feature extraction through feature selection, and performance comparison. Multiple classifiers are created using this process. Feature selection algorithms are compared using the data subsets each creates. Multiple data sets and models are created, and the performances of these classifiers are compared and discussed
DEEP LEARNING FOR COMPUTER VISION APPLICATIONS IN MEDICAL DIAGNOSTICS AND WILDLIFE MONITORING
Get PDFThis dissertation explores innovative applications of deep learning and computer vision techniques across three distinct domains: medical imaging, dermatological diagnostics, and wildlife monitoring. The research addresses critical challenges in each field through the development and optimization of convolutional neural networks and other deep learning architectures.
The first study examines COVID-19 classification from X-ray images, comparing one-shot versus two-stage classification approaches using transfer learning with pre-trained models such as VGG16 and VGG19. The initial hypothesis was that breaking down the classification task into two optimized tasks would yield better results than one-shot classification. Results demonstrated that the single-stage approach achieved superior performance with 95% accuracy, contrary to initial hypotheses. The second study tackles the pervasive problem of class imbalance in dermoscopic image classification for skin cancer detection. Two approaches were systematically compared: traditional data augmentation techniques and Generative Adversarial Networks (GANs) for synthetic image generation. Fine-tuned deep learning models, including EfficientNet, ResNet50, Vision Transformers, and ConvNeXt, were evaluated on the augmented dataset and the synthetic dataset. Data augmentation proved more effective than GAN-generated synthetic images, with EfficientNet achieving 97.7% accuracy. The third study extends deep learning techniques to wildlife monitoring and conservation, developing a system for detecting and tracking beluga whales in aerial drone footage. The YOLOv7 object detection model achieved high precision and recall (92%–92% for adult belugas, 94%–89% for calves), while a novel post-processing algorithm improved multiple objects tracking accuracy from approximately 30% to 70%.
The collective findings advance the state of the art in applying deep learning to visual data across diverse domains, demonstrating effective solutions to key challenges in classification, class imbalance, and object tracking. This research contributes practical approaches that enhance diagnostic capabilities in healthcare, improve dermatological screenings, and support wildlife conservation efforts through more efficient monitoring techniques
The Use of AI Tools for Enhancing Digital Library Services With Information Architects as Responsible Partners
No full textDigital Libraries Across Continents illustrates how digital librarianship practitioners and scholars digitize, exhibit, and preserve their cultural heritage, and how these practices may be influenced by the policy, economic, and sociocultural environments in which they are developed.
Including scholarly articles, case studies, examples of best practice, and conceptual essays solicited from different continents, this book provides an overview of the status quo of digital libraries around the globe. The case studies examine how macro-level policy, funding, and social priorities influence the development of digital libraries. The volume offers a deeper understanding of the similarities and differences between libraries in different countries and the ways in which they view, foster, develop, and sustain digital librarianship. Chapters within the book examine systems, standards, workflows, content, protocol, social and policy environments, culture, metadata, and more, through a series of case studies provided by practitioners working in these settings. Taking a comparative international approach, the book promotes the development of inclusive, accessible, and sustainable digital libraries that serve a global human knowledge endeavor.
Digital Libraries Across Continents provides a wide-ranging examination of issues in cross-border digital library contexts. It will be essential reading for library practitioners, as well as information scientists and educators
ANTHROPOGENIC AND NATURAL INFLUENCES ON BEACH GEOMORPHOLOGY AND COASTAL RESILIENCY
Get PDFBeaches are complex, dynamic systems that provide important habitat, protection from storm impacts, support tourism, and provide numerous benefits to the economy. Coastal erosion is a natural phenomenon but is becoming a growing problem due to factors such as rising sea levels, intensity of major storms, lack of sand input from rivers, and anthropogenic activities. Florida\u27s 825 miles of sandy beaches are experiencing accelerated erosion and adaptive management practices are needed to ensure sufficient sediment remains in the littoral system. Beach geomorphology and sedimentology research is critical to understand complex coastal processes to improve coastal management strategies for the resiliency of sandy beaches.
This dissertation addresses important research gaps for coastal morphodynamics and beach management through three research objectives presented in three separate chapters. Chapter 2 evaluates the effectiveness of proxy-based shoreline methods over a 17-year period to assess their reliability for estimating shoreline and beach volume changes. The third chapter examines the influence of beach morphology and sediment characteristics on sea turtle nesting, hatching, and emergence success across managed and non-managed beaches. The last chapter examines the impacts of extreme storm events on beach morphology and property damage, integrating socio-demographic data to inform resilience planning.
It was determined that proxy-based shoreline methods can be effective over long timescales to estimate beach volume along the south Florida coast. However, shoreline proxies underestimate beach volume at short-time scales, particularly in urbanized areas. Storm impacts resulted in complex relationships between shoreline change and volume change, influenced by barrier morphology, structure type, and coastal features like saltwater marshes. Sea turtle nesting success was found to be strongly influenced by sediment grain size, sorting, and temperature, with higher success rates occurring on mixed-management south Florida beaches comprised of cooler, finer, and more well-sorted sediments. Lastly, the novel methodology conducted to model ebb surge water channelization showed that a higher number of damaged structures occurred within proximity of the channels along a low-lying barrier island in southwest Florida. The findings from this dissertation highlight the need for site-specific, data-driven coastal management approaches to improve resilience measures for the future. As coastal populations and infrastructure continue to grow, understanding these dynamics will be critical for long-term coastal sustainability and protection
CAN SERVANT LEADERSHIP REINVIGORATE COMMUNITIES OF FAITH
Get PDFThis study examined the relationship between servant leadership characteristics and church growth in Episcopal congregations. Using quantitative analysis, the research assessed whether leadership characteristics—such as emotional healing, creating value for the community, conceptual skills, empowering followers, helping followers succeed, putting followers first, and behaving ethically—were significant predictors of numerical growth. Data were collected from 40 churches (20 growing, 20 declining), with leadership perceptions measured through a validated servant leadership instrument.
The findings indicate that while growing churches exhibited slightly higher mean scores across several servant leadership characteristics, none of the differences were statistically significant. Logistic regression analysis further demonstrated that no individual leadership characteristic significantly predicted church growth at the p \u3c .05 level. These results suggest that church expansion may be influenced more by external factors—such as congregational engagement, denominational policies, and regional demographics—than by leadership characteristics alone.
This study contributes to church leadership research and servant leadership theory by highlighting the complexity of leadership effectiveness in faith-based organizations. The findings reinforce the need for a context-dependent approach to leadership, where contextual, cultural, and structural factors are considered alongside leadership behaviors. Given the lack of statistically significant findings, future research should incorporate qualitative methods to explore how servant leadership manifests in different congregational settings and whether leadership practices align with broader church growth strategies
ENHANCING THERMAL AND MECHANICAL PROPERTIES OF COUNTERTOP EPOXY THROUGH 2 NANOPARTICLE REINFORCEMENT
No full textCountertop epoxy is a popular material for kitchen, bar, and commercial surfaces because of its glossy outlook, heat resistance, and longevity. However, its performance under severe conditions is hampered by its low thermal stability and intrinsic brittleness. The goal of this investigation is to improve tabletop epoxy\u27s mechanical and thermal characteristics by adding zirconia (ZrO₂) nanoparticles. Expanding upon earlier tests with different fillers, 2 wt% Mg₂Al(OH)₆(CO₃)₀.₅·nH₂O (LDH) nanoparticles initially showed moderate performance improvement. However, comparative analysis revealed that ZrO₂ outperformed LDH across most key metrics. Comprehensive testing—including Differential Scanning Calorimetry (DSC), tensile, flexural, and impact analyses— demonstrated that 2 wt% ZrO₂ increased the glass transition temperature (Tg) from 56 °C to 70 °C, indicating a 25% gain in thermal stability. Meanwhile, 1 wt% ZrO₂ yielded the highest mechanical enhancement, with a 70% increase in flexural modulus and significant improvements in tensile strength and impact resistance. To further explore synergistic effects, a hybrid nanofiller system was synthesized by combining 1 wt% ZrO₂ with 0.5 wt% LDH in a 2:1 ratio (1.5 wt%). Although this hybrid configuration was anticipated to balance toughness and thermal stability, the results demonstrated otherwise. The hybrid composite showed a 27% decrease in tensile strength and a 54% reduction in Young’s modulus compared to neat epoxy, indicating that the combination did not yield favorable reinforcement. Although the blended composite showed a few gains, it was not able to outperform ZrO₂ alone. According to these results, it is evident that 1 wt% ZrO₂ is a very efficient reinforcement material that provides an excellent balance between mechanical and thermal endurance for countertop epoxy. In addition, the nanocomposites significantly increase the glossiness
THE CONSISTENCY STRENGTH OF THE GENERALIZED CONTINUUM HYPOTHESIS FAILING AT A MEASURABLE CARDINAL
Get PDFThis expository paper investigates the equiconsistency of the GCH failing at a measurable cardinal with the existence of a cardinal κ of Mitchell order κ++.
The upper bound of this equiconsistency follows in two parts: Assuming the existence of a model of o(κ) = κ++, one can first use an argument of Gitik to force the existence of an elementary embedding satisfying certain closure conditions, then use a forcing due to Woodin to force the failure of GCH at κ while preserving the measurability of κ. It is this Woodin result which this thesis focuses on in the upper bound.
The lower bound of this equiconsistency is an inner-model-theoretic argument due to Mitchell, where one can show that assuming the GCH fails at a measurable cardinal, then K, the so-called ‘core model below o(κ) = κ++’ exists. This thesis aims to bridge a gap in the literature by providing a much-needed approachable introduction to inner model theory at the level of o(κ) = κ++ for the non-specialist. Mitchell’s argument that the GCH failing at a measurable cardinal implying the existence of a model of o(κ) = κ++ is then given
HISTORICAL CONTEXT IN MATHEMATICS EDUCATION: IMPROVING STUDENT ENGAGEMENT AND ACHIEVEMENT
No full textThere are many informational gaps in contemporary mathematics education. Students see the mathematical topics they are being taught as facts and procedures unrelated to each other or to themselves as mathematicians. Many ancient developments in the realm of mathematics are still relevant today, showcasing the needs that motivated their longstanding evolution. By including the stories behind the formulas in mathematics classrooms, students from diverse backgrounds can feel seen and connected to their education. The inclusion of the holistic background of mathematics cannot be nurtured in the classroom without buy-in on the parts of both students and instructors, as well as the education of the instructors so that they are able to successfully teach mathematics in a way that showcases the true value of the material
DESIGN AND SYNTHESIS OF PEPTIDE BASED SYSTEMS OF INTRANASAL RNA DELIVERY
No full textA promising strategy that could be used to treat many central nervous system (CNS) diseases, including Parkinson’s disease, Alzheimer’s disease and brain cancer, is through gene silencing by short interfering RNAs (siRNA). siRNA induces gene silencing by targeting complementary mRNA for degradation. However, a key challenge in the development of siRNA-based therapies for the CNS diseases is their transport across the blood–brain barrier (BBB) and blood-cerebrospinal fluid barrier (BCSFB). To facilitate the transport of siRNA across the BBB and BCSFB we have designed a siRNA delivery system based on the 17-mer naturally occurring cyclic peptide odorranalectin (OL) carrying multiple positive charges. Our group has shown previously that OL preferentially binds to the L-fucose and to a lesser degree D-galactose and N-acetyl-D-galactosamine, which are widely distributed on the olfactory epithelium of nasal mucosa, suggesting a possibility for intranasal delivery. As a proof of principle, we have prepared a series of positively charged OL-based peptides. This was achieved by incorporating poly arginine sequences (poly-R) of various lengths, as well as the sequence of the TAT peptide (YGRKKRRQRRR), derived from the human immunodeficiency virus TAT protein, into the OL scaffold. We hypothesized that the cationic sequence of OL will bind negatively charged siRNA, while the carbohydrate-binding sequence of OL will facilitate nose-to-brain transport of the siRNA/cationic-OL complex. The synthesized cationic peptides were tested for SiRNA binding using Circular Dichroism (CD) spectroscopy, and gel electrophoresis, while Isothermal Titration Calorimetry (ITC) was used to evaluate if addition of cationic peptides to OL N-terminal side doesn’t hinder the ability of OL to bind glycoproteins
HUMERAL HEAD RETROVERSION ON A CONTEMPORARY SKELETAL POPULATION WITH KNOWN PHYSICAL ACTIVITY TO INVESTIGATE THE DIFFERENTIAL EXPRESSIONS OF BIOLOGICAL SEX
No full textRetroversion is an angled measurement of the humerus that is indicative of habitual overhand activity. Elevated retroversion measurements (in degrees) indicate overhand activity. Overhand activity indicates subsistence gathering from spear or atlatl throwing. Most retroversion research is conducted in populations where men perform more physically demanding labor. Therefore, bioarchaeologists are producing the same data over and over. Additionally, contemporary retroversion data is used as comparative figures. These subjects range from youth to professional baseball players, but the data on women is nearly obsolete. Bioarchaeology needs contemporary female retroversion data. Research was conducted at the University of Tennessee Knoxville where humeral head retroversion was measured on both left and right humeri in populations of known physical activity to investigate how retroversion materializes on the skeletal body between sexes. Data analysis was conducted using Jupyter Notebook. Left-right asymmetry can be used as a new model or method in predicting overhand activity in women