19304 research outputs found
Sort by
Development of Piezoelectric BaTiO₃/PVDF Nanocomposite Films for Sensing and Energy Harvesting Applications
No full textPiezoelectric polymer nanocomposites have garnered significant interest due to their lightweight, flexibility, and ability to convert mechanical energy into electrical energy, making them ideal for various applications, such as sensors, energy harvesting devices, and actuators. Among these materials, polyvinylidene fluoride (PVDF) and its copolymers are particularly notable for their piezoelectric properties, which arise primarily from the presence of the electroactive β-phase. Therefore, enhancing the β-phase content in PVDF is crucial for optimizing its piezoelectric performance. This dissertation begins by focusing on optimizing the β-phase content in PVDF through the careful optimization of experimental parameters, including solvent choice, solution concentration, sonication time, and crystallization and annealing temperatures. Barium titanate (BTO) nanoparticles, a ceramic material with strong piezoelectric characteristics, were synthesized and incorporated into the optimized PVDF matrices to create BTO/PVDF nanocomposites. In the first approach, a BTO/PVDF layer was spin-coated onto an aluminum nitride (AlN) layer, a material known for its excellent piezoelectric properties, with the aim of improving piezoelectric performance by minimizing the leakage current typically observed in sputtered AlN films. In the second approach, electrospinning was employed to fabricate BTO/PVDF nanofiber films as a standalone platform. Various types of BTO nanoparticles, including cubic, tetragonal, PDA-coated, fluorosilane-functionalized were embedded to evaluate their effects on the piezoelectric performance. Both approaches show promising results in enhancing the β-phase content of PVDF and enhancing the overall piezoelectric performance of the nanocomposite. The combination of BTO, PVDF, and AlN in thin films, along with the electrospun BTO/PVDF nanofibers, offers a pathway toward developing highly efficient and flexible piezoelectric materials suitable for sensing and energy harvesting applications
Unsupervised Learning via Autoencoded UMAP-Enhanced Clustering
No full textUnsupervised learning is a crucially important field in machine learning. One of its main tools is clustering, a technique used in data analysis to identify groups of similar objects within a dataset. It involves partitioning the data into subsets, or clusters, where data points within the same cluster are more similar to each other than to those in other clusters. However, clustering performance and thus the quality of unsupervised learning depends greatly on a particular choice of the measure of proximity of points in the data space. Here we propose to learn such a measure concurrently with the clustering task itself. This is achieved by learning a non-linear embedding of the data into a low-dimensional space, where the proximity of data points to each other can be measured in Euclidean sense. The embedding is performed via a composition of two mappings, the first being the encoder of an autoencoder while the second is the Uniform Manifold Approximation and Projection (UMAP). The autoencoder is trained with a loss that includes both the conventional data reconstruction component as well as a clustering-promoting one. We combine both embeddings and the subsequent clustering into a single three-stage unsupervised learning framework referred to as Autoencoded UMAP-Enhanced Clustering (AUEC). We evaluate the performance of AUEC on the MNIST dataset employing a convolutional neural net as the autoencoder. The accuracy achieved by the proposed approach on MNIST data rivals that of the state-of-the-art unsupervised learning techniques
Advancing Community Detection through Ensemble Learning and Modularity Maximization
No full textArguably, the most fundamental problem in Network Science is finding structure within a complex network. Often, this is done by partitioning the network's nodes into communities in a way that maximizes an objective function. However, finding the maximizing partition is generally a computationally difficult NP-complete problem. Recently, a machine-learning algorithmic scheme was introduced that uses information within a set of partitions to find a new partition that better maximizes an objective function. The scheme, known as RenEEL, uses extremal ensemble learning. Starting with an ensemble of partitions, it updates the ensemble by considering replacing its worst member with the best of partitions found by analyzing a reduced network formed by collapsing nodes, which all the ensemble partitions agree should be grouped together, into super-nodes. The updating continues until consensus is achieved within the ensemble about what the best partition is. The original ensemble partitions and each of the partitions which is used for the update, are found using a simple ``base" partitioning algorithm. We conduct an empirical study of RenEEL’s effectiveness as a function of ensemble parameters and relate the results to extreme value statistics. It shows that increasing the ensemble size yields better results in approaching ensemble size . Building on this foundation, we extend the RenEEL framework to the domain of bipartite networks, where community detection presents the unique challenge of the resolution limit in modularity-based methods. We first demonstrate how a benchmark bipartite network fails to resolve smaller communities using traditional modularity. We then introduce a new metric, Generalized Bipartite Modularity Density (), which leverages the resolution limit to reveal hierarchical community structures in bipartite networks by allowing tunable control over resolution. This consistently outperforms existing metrics in detecting communities in both benchmark and real bipartite networks. Using RenEEL to maximize this novel metric, we demonstrate its effectiveness in uncovering hierarchical community structures across a range of real-world bipartite networks, including the Southern Women network, an Asthma-Patient bipartite network, and a Psychological Item–Embedding Bipartite Network. Applying this to the Southern Women network, we show that for lower values of , recovers the canonical partition reported in earlier studies. As increases, it uncovers progressively finer subgroups, ultimately revealing a richer hierarchy of smaller clusters that together provide, for the first time, a unified view of the network’s multi-scale social organization. In the Asthma-Patient bipartite network of patients and cytokines, recovers known baseline groupings at standard resolution. As resolution increases, it reveals finer modules linking specific cytokines to patient subsets, suggesting potential therapeutic targets. Notably, some slightly different groupings from the standard partition also emerge, highlighting structural features that may inform asthma treatment. In the Psychological Item–Embedding Bipartite Network derived from an LLM fine-tuned on annotated survey data, effectively captures nuanced relationships between survey items and latent model dimensions, revealing meaningful clusters that align with conceptual similarities learned by the model. These results highlight that RenEEL, when combined with problem-specific metrics such as , offers a robust and generalizable approach for detecting multi-scale community organization in diverse bipartite systems
A Diverse and Comprehensive Air Quality Modeling Analysis of Houston, Texas: How Will Changing Emissions, Industries, and Legislation Impact the Air and Human Health?
No full textThis dissertation utilizes air quality models to assess drivers of pollution in Houston, Texas, where air quality poses a significant risk to human health. This research focuses on how factors like emissions from the energy sector, Houston Ship Channel (HSC) activity, and evolving legislation impact the formation of ozone (O3) precursors (such as volatile organic compounds (VOCs) and nitrogen oxides (NO + NO2 = NOx)), in addition to particulate matter (PM2.5 and PM10) and greenhouse gases. Houston’s continuous growth, which intensifies transportation and energy demands, further poor air quality. To address these challenges, this dissertation provides a three-part investigation to inform more effective air quality control. Chapter 1 compares surface O3 formation in Houston’s urban and industrial environments using a novel two-step approach that combines Positive Matrix Factorization (PMF) with Random Forest machine learning interpreted by SHapley Additive exPlanation (SHAP). VOC and NOx data from the urban Milby Park and industrial Lynchburg Ferry sites were analyzed for the O3 seasons of 2017-2021. This revealed that NOx emissions suppress O3 formation in the NOx-saturated urban core while promoting formation at the industrial sites. Chapter 2 explores the connection between Houston’s air quality legislation and emissions. A comprehensive legislative timeline was constructed from 1952 to 2023, which demonstrates landmark legislation like the Clean Air Act (CAA) and the Texas Emissions Reduction Plan (TERP). The joint PMF-AI approach was applied to air quality data from 1990-2000 and 2017-2021 for comparison at the Clinton and Deer Park #2 sites to assess the impact of legislation on model results. Analysis revealed that emissions source profiles evolved over time, with decreased NOx contribution, as well as key drivers of O3 formation like temperature. Chapter 3 provides a multidecadal analysis of Houston’s pollution drivers by applying the PMF-AI approach to a 30-year dataset (1990-2021) from Clinton and Deer Park #2. By analyzing time intervals during this period, this chapter provides long-term analysis of emissions sources and predicted ozone formation. Houston showed an evolution from industrial-dominant sources in the 1990s to a more complex mix more recently; knowing this allows for more effective legislation implementation in the future
Crustal, Thermal, and Structural Controls on Hydrocarbon Systems in the Delaware Basin, West Texas, and Southeastern New Mexico
No full textThis dissertation presents an integrated analysis of the geophysical and geothermal expression of Mesoproterozoic – early Miocene tectonic events affecting hydrocarbon production in the Delaware Basin, the western physiographical province of the complexly structured, intracratonic Greater Permian Basin based on the interpretation of gravity and magnetic data, 3D seismic, and well data. Initial ambiguity of petroleum systems modeling results led to the discovery of unconstrained thermal and tectonic parameters within the Delaware Basin. Each chapter in this study contributes new constraints on the thermal and tectonic history of the basin and develops new and repeatable associated methodologies. Chapter 2 uses gravity and magnetic data and structural mapping based on the stratigraphic correlation of 27 horizons using 11,433 formation top picks from 1,181 wells constrained by pre-stack depth migrated (PSDM) 3D seismic interpretation to address the influence of Proterozoic basement on interval-specific geothermal gradient calculations and Wolfcampian production trends within the Delaware Basin. The study highlights the importance of empirically derived temperature-depth functions revealing significant variability in the geothermal profiles due to: 1) upper crustal lithology; 2) proximity to basement-rooted fault systems; and 3) burial of the western basin by late Cretaceous Laramide strata. I interpret lateral temperature variation within the Wolfcampian–Precambrian interval to be influenced by the radiogenic heat production (RHP) of underlying igneous intrusive complexes and northeast-southwest-trending crustal terranes associated with Paleo-Neo-Proterozoic accretion and collision processes. Chapter 3 explores the seismic characterization and structural modeling of shallow extensional fault systems impacting hydrocarbon production within the west-central Delaware Basin using multi-layer 3D seismic attribute analysis, fault framework modeling, and well data to establish a correlation between shallow lineaments, drilling complications, elevated water-oil ratios (WORs), and excessive H2S production. I establish a repeatable seismic workflow for 3D imaging and subsequent structural modeling of 60 previously undocumented faults along 27 newly mapped features within the southwestern Delaware Basin and constrain the timing of regional extension and exhumation to the middle Oligocene-early Miocene in response to Basin and Range tectonism and Rio Grande rifting
The Role of Cognition and Sociodemographics on Health Brokering Performance in a Diverse Adult Sample
No full textBackground: Health brokering is a phenomenon through which individuals assist their loved ones (usually older and/or with lower literacy levels) in understanding health-related subjects, or engaging with healthcare services, by acting as intermediaries and interpreters of information on their behalf. Health brokering may be associated with key demographic and clinical factors closely known to drive disparities in health care, serving as a mechanism for underserved communities to leverage shared knowledge and advocate for efficient health services. However, little is known about the predictors of successful health brokering; understanding the factors that enhance or limit health brokering performance may be key to contextualizing the health care experiences of diverse patients, particularly those at higher risk for chronic conditions. Objectives: The purpose of the study was to determine the relationship between health brokering performance (see Appendix A) and the following outcomes of interest: sociodemographic and cultural measures, health literacy, cognitive and physical functioning, and mental health. In addition, this study expanded the existing literature on health brokering by relying on a novel, performance-based health brokering measure in addition to existing self-report tools. Method: A total sample of 96 undergraduate students from the University of Houston and 48 community community-dwelling adults were recruited to participate in this study. Undergraduate students and community participants all completed a two-stage study consisting of an online survey and an in-person assessment. Correlation matrices were used to address Aim 1, while linear regression analyses were used to address the goals described in Aims 2 through 4. Results: Sociodemographic and cultural factors—such as socioeconomic status, family relationship quality, and alignment between individual and familial cultural identity—were modestly associated with performance on a novel health brokering task. Among health literacy measures, only eHealth literacy significantly predicted brokering performance, highlighting the relevance of digital navigation skills. Cognitive performance emerged as a robust predictor, particularly among younger participants, while physical and emotional health had limited explanatory power. In a hierarchical model, executive functioning accounted for the greatest proportion of variance in brokering performance, after controlling for demographic and psychosocial factors. Discussion: Findings suggest that health brokering is shaped by a combination of social positioning and individual cognitive abilities. The predictive value of eHealth literacy emphasizes the growing role of digital competencies in health navigation. Results highlighting cognition as a key predictor of brokering performance are consistent with the notion that brokering involves high-level processing and decision-making. Notably, differences across participant groups further suggest that life experience may moderate how these skills are deployed. Conclusion: Health brokering competence appears to stem from a complex interplay of cognitive, cultural, and relational factors. Digital and executive functioning skills, in particular, play a central role in effective brokering, while demographic and relational contexts shape the sociocultural experience of undertaking health brokering roles. These insights point to the value of supporting health brokers through institutional practices in order to meet the real-world demands of diverse patient populations
Preserving Visionary Art Environments: The Orange Show, Stewardship, and Intent
No full textThis thesis examines the Orange Show in Houston, Texas, as a case study in the conservation of visionary art environments. Constructed by postman Jeff McKissack over several decades, the Orange Show is a space that defies conventional art and architectural categorization, and its preservation challenges standard conservation practices. This work positions the site as a living system shaped by social, ecological, and institutional forces. Through archival research, interviews, and site analysis, the study argues for an adaptive, participatory model of preservation where continuity of meaning and the value of intangible heritage is privileged over material authenticity. The Orange Show becomes a lens through which to rethink preservation ethics, embracing impermanence, communal stewardship, and the evolving life of unconventional cultural spaces
A Comparison of Methods to Quantify Nano- and/or Microplastic (NMPs) Deposition in Wild-Caught Eastern Oysters (Crassostrea virginica) Growing in a Heavily Urbanized, Subtropical Estuary (Galveston Bay, USA)
No full textNano- and microplastics (NMPs) in waterways reflect the impact of anthropogenic activities. This study examined spatial variations in the presence and types of NMPs in Galveston Bay (Texas, USA) surface waters and eastern oysters (<i>Crassostrea virginica</i>). The results reveal most MPs carried by surface waters are fibers &gt; films &gt; fragments. Up to 200 MPs were present in individual oysters [=1.88 (&plusmn; 0.22 SE) per g wet weight]. Oyster health, based on condition index, varied spatially, but was not correlated with MP load. Based on attenuated total reflectance&mdash;Fourier-transform infrared spectroscopy, polyamide and polypropylene were frequently found in waters in the upper bay while ethylene propylene and polyethylene terephthalate were more common in the lower parts of the bay. Pyrolysis&ndash;gas chromatography&ndash;mass spectrometry revealed a very large range in concentrations of NMPs, from 28 to 10,925 &micro;g &sum;NMP/g wet weight (or 172 to 67,783 &micro;g &sum;NMP/g dry weight) in oysters. This chemical analysis revealed four main types of plastics present in oysters regardless of location: polypropylene, nylon 66, polyethylene and styrene butadiene rubber. Based on this finding, the average daily intake of NMPs estimated for adult humans is 0.85 &plusmn; 0.45 mg NMPs/Kg of body weight/day or a yearly intake of 310 &plusmn; 164 mg NMPs/Kg of body weight/year. These findings reveal higher body burdens of plastics in oysters are revealed by the chemical analysis relative to the traditional approach; this is not unexpected given the higher sensitivity and selectivity of mass spectrometry and inclusion of the nanoplastic particle range (i.e., &lt;1 mm) in the sample preparation and analysis
Exploring the Relationships between Acculturation, Somatization, and Problematic Alcohol Use: A Moderation Analysis of South Asian Americans
No full textBackground: High rates of alcohol consumption are associated with numerous adverse physiological, somatic, and psychological health implications. Previous studies have highlighted patterns of increased alcohol consumption associated with acculturation in the United States (US), likely related to the influence of cultural drinking norms. A large portion of the extant literature has focused on Hispanic American populations or broadly on the Asian American US population. Less is known about alcohol use patterns amongst specific Asian American subgroups. South Asian Americans are one such understudied group, despite their status as the most rapidly expanding segment of the US Asian American population. South Asians living in the US have been observed to be at higher risk of binge drinking relative to other Asian American subgroups. South Asian Americans also experience higher rates of somatization, or physical manifestations of psychological distress, than their Western counterparts. Considering the impacts of acculturation on alcohol use and South Asians’ unique cultural patterns of somatization, additional research is needed to examine the relationships between these constructs. Purpose: The present study aimed to examine whether somatization moderates the relationship between acculturation and alcohol use problems in a sample of South Asian adults living in the US. Methods: A sample of 220 South Asian American adults completed an online survey related to alcohol use, acculturation, and somatic symptoms. These data were used to test the proposed moderation model using ordinary least squares (OLS) regression. Results: Study results indicated significant main effects of Western acculturation and somatization on participants’ alcohol use patterns, as well as a significant moderating effect of somatization on the relationship between Western acculturation and alcohol use. Conclusion: The current findings were unanticipated and challenge the prevailing literature relating to the influence of Western acculturation on alcohol consumption patterns. Nevertheless, somatization was found to be a particularly critical variable in understanding substance use patterns amongst South Asian Americans. Findings of the present study can be used to inform clinical prevention, screening, and treatment efforts aimed at South Asian American communities
It’s About Perspective: An Analysis of First Vs Third Person Narrative in a Health Campaign About Postural Orthostatic Tachycardia Syndrome
No full textThis study explored how message format (video vs. text) and narrative perspective (first-person vs. third-person) affect how people respond to health campaigns about invisible illnesses. The goal was to better understand which combinations help people feel more connected to the message and more likely to support those with invisible conditions like POTS. Participants (N = 106) were randomly assigned to one of four versions of a health story. After viewing or reading the message, they answered questions about how immersed they felt (transportation), how much they related to the main character (identification), whether they thought about their own life while engaging with the message (self-referencing), how they felt about the message, and how likely they were to take action. The results showed that first-person stories led to higher transportation and more self-referencing than third-person ones. However, narrative perspective didn’t significantly affect identification, message perception, or behavioral intention. Video didn’t significantly increase behavioral intention either, but it did show a trend in that direction. Participants who watched a video were slightly more likely to report wanting to help people with invisible illnesses than those who read text. No significant interaction was found between modality and perspective, meaning one didn’t boost or reduce the effects of the other. These findings suggest that using first-person stories may help audiences feel more immersed and reflect more personally on the message. While not every effect was significant, the patterns support the idea that small differences in how a message is told can shape how people respond. This adds to existing research on health campaigns and shows the value of personal storytelling in raising awareness for conditions that aren’t always visible