19304 research outputs found
Sort by
Quake-OVUDA: Component-Level AI-Based Post-Earthquake Building Inspections with Vision–Language Guided Unsupervised Domain Adaptation
No full textThe semantic segmentation of building components and their damage states is a critical step toward automating structural inspections in the aftermath of disasters. A key challenge towards training deep neural networks for these tasks is the manual development of large, annotated datasets for the real world. Training Unsupervised Domain Adaptation (UDA) models, using synthetic, automatically generated imagery, has shown promise in reducing reliance on manual labeling and improving the adaptation of features learned from the synthetic domain to the real-world domain of images. However, existing UDA methods can still struggle to generalize to certain real-world representations of objects that are stylistically different between synthetic and real domains or are generally underrepresented in the synthetic dataset. Domain gaps like these are especially relevant in synthetic structural imagery, where real-world building components and damage can adopt a wide variety of patterns. To address these limitations, this thesis incorporates features learned from open-vocabulary (OV) semantic segmentation into the UDA framework, DAFormer, thereby enabling improved contextual understanding and transferring of features from the synthetic to real domains. Specifically, the Simple Encoder Decoder (SED) OV model is fine-tuned on the synthetic dataset, QuakeCity, and is used to generate pseudo-labels for unlabeled real-world images that are used in DAFormer’s self-supervised learning step. To further improve classification performance, particularly of nuanced damage classes, additional refinement and classification of instances predicted by Quake-OVUDA are performed by leveraging the vast context provided by vision language models (VLMs). The proposed framework, termed Quake-OVUDA, and its additional supplementation with VLM expertise, outperform direct segmentation, open-vocabulary segmentation, and baseline UDA methods for building component and component damage state segmentation tasks. Evaluation of the proposed methods shows improvements of up to 5.49 mean Intersection-over-Union (mIoU) for the component segmentation task and up to 11.02 mIoU for the damage state segmentation task over the baseline UDA approach. This work demonstrates an effective combination of OV segmentation and VLM-based reasoning to guide UDA and reduce reliance on manually annotated real-world data for visual building assessment
Advancing Air Pollution Forecasting and Health Risk Assessments: A Synergistic Approach Integrating Deep Learning Models for Health and Socioeconomic Evaluations
No full textAir pollution continues to pose significant threats to public health and the environment. This dissertation advances air quality forecasting and health impact assessment by integrating artificial intelligence, deep learning, and epidemiological modeling. First, an Artificial Neural Network (ANN) model was developed to forecast the Air Quality Index (AQI) and PM₁₀ concentrations in Tehran, Iran, incorporating traffic data, green space proximity, meteorological conditions, and temporal factors. The model achieved high predictive accuracy (R = 0.82 for AQI and R = 0.93 for PM₁₀). Health impact assessment using forecasted PM₁₀ values indicated that 6.87% of respiratory symptoms among children and 9.72% of chronic bronchitis cases among adults were attributable to elevated PM₁₀ exposure. Second, three deep learning models—Deep Convolutional Neural Networks (Deep-CNN), Long Short-Term Memory (LSTM) networks, and Deep Neural Networks (DNN)—were evaluated for forecasting daily maximum ozone (O₃) across 19 provinces in South Korea. Deep-CNN achieved the highest accuracy (IOA = 0.93 on forecast day 1). Significant correlations were found between O3 and female respiratory mortality (r = 0.53, ρ = 0.42; p = 0.020, 0.024), cardiovascular mortality in both genders, and male employment (r = 0.48, ρ = 0.76; p = 0.039, 0.0002). Female employment showed weaker linear correlation (r = 0.42, p = 0.061), but a strong monotonic trend (ρ = 0.74, p = 0.0003). Third, a novel Deep Bias Correction (Deep-BC) framework using CNNs was introduced to correct biases in CMAQ O₃ forecasts. Applying Deep-BC’s improved forecasts to AirQ+ modeling showed that short-term O₃ exposure accounted for 0.40–0.48% of deaths from natural causes and respiratory diseases, and 0.67–0.81% of cardiovascular deaths across major South Korean provinces, with higher burdens among men. Together, these studies demonstrate that machine learning–enhanced forecasting, combined with health and socioeconomic analysis, offers powerful tools for mitigating the impacts of air pollution and supporting public health and socioeconomic equity
Equation of State Based on Lattice QCD: Expansion Scheme and Critical Point
No full textExploring the Quantum Chromodynamics (QCD) phase diagram at finite baryon density is essential for understanding the behavior of strongly interacting matter. QCD, the fundamental theory of the strong interaction, predicts a transition from hadronic matter to a deconfined quark-gluon plasma (QGP), an exotic phase believed to have existed in the early universe and potentially persisting in neutron star cores. While this phase is routinely recreated in heavy-ion collisions at RHIC and the LHC, the precise nature of the QCD transition remains unresolved. At vanishing baryon chemical potential, lattice QCD confirms a smooth crossover; however, the existence and location of a first-order phase transition and a critical point at finite density remain elusive due to the sign problem, which hinders direct lattice simulations. This dissertation addresses the challenge of constructing reliable QCD equations of state (EoS) at finite baryon density by applying and investigating a novel extrapolation scheme that improves upon traditional Taylor expansions. The method enables a stable extension to larger baryon chemical potentials, although its reliability remains under investigation. To study its convergence properties, the high-order coefficients of this expansion are computed using effective models such as the Hadron Resonance Gas and Cluster Expansion Model. The study found that this expansion scheme exhibits superior convergence properties compared to Taylor series, particularly near the critical region. Building on this foundation, I introduce a critical point into the QCD EoS using the universality class of the three-dimensional Ising model. A tunable mapping scheme embeds the critical behavior within the QCD phase diagram, ensuring thermodynamic consistency and causality. The resulting EoS captures a comprehensive set of thermodynamic quantities across a broad range of parameters and is made available through open-source software developed within the MUSES Collaboration. Furthermore, this work generalizes the expansion scheme to a four-dimensional framework, resulting in a 4D EoS that involves the independent control of temperature and three conserved charges (baryon number, strangeness, and electric charge). Using continuum-extrapolated lattice QCD data, this generalization significantly expands access to the QCD phase diagram, enabling extrapolation to high densities. Together, these developments provide a robust theoretical framework for studying QCD matter at finite density, with significant implications for interpreting heavy-ion collision data and constraining the location of the QCD critical point. The findings also lay the groundwork for future integration with hydrodynamic simulations and Bayesian inference techniques, facilitating direct comparisons with experimental observables and enabling a data-driven extraction of QCD EoS parameters
An Investigation of the Color Change in LuH2 at Megabar Pressures
No full textThe claim of room temperature superconductivity (RTS) in a nitrogen doped lutetium hydride sample (LuHxNy) at the incredibly low pressure of ~ 1 GPa with an equally incredible superconducting transition temperature (Tc) of ~ 294 K sent the physics community and science-news sources in general into a frenzy earlier in 2023. Coupled with this breakthrough in conductivity was an interesting color-change in the LuHxNy samples. With the addition of pressure, the material changed color at room temperature, from: a blue non-superconducting phase (ambient – 0.3 GPa), through a pink superconducting phase (0.3 – 3 GPa), and finally to a red non-superconducting metallic phase for pressures above 3 GPa. Hence, the focus of this study investigates the temperature dependence of resistance and color changes in lutetium dihydride (LuH2) under pressure up to 116 GPa. The samples were characterized at ambient pressure through powder X-ray diffraction (pXRD) and upon loading into the diamond anvil cell (DAC) the pressure-induced color changes (blue-violet-pink-red) reported by the Dias' group were reproduced. Alongside many others who worked in the Lu-H-N system, no evidence of superconductivity was found in our LuH2 sample between 3.1 GPa and 116 GPa from resistance measurements. Despite this, a high-pressure grey phase transition starting near 50 GPa which persists up to 116 GPa was observed. Optical microscopy measurements were coupled with the Red-Green-Blue (RGB) and Hue-Saturation-Lightness (HSL) color scales to quantify and compare an array of studies to see the effect of composition and pressure on the Lu-H-N system. We then correlate these findings to electronic properties through resistive measurements under pressure
Effects of Acute Aerobic Exercise and Exercise Intensity on Metabolism and Function of CD8+ T Cells
No full textAcute exercise affects both systemic energy metabolism and CD8+ T cell (CD8+) distribution and function. Importantly, these effects are dependent on exercise intensity. Furthermore, striking similarities exist between the systemic metabolic effects of exercise and the role of energy metabolism in CD8+. However, how exercise affects CD8+ metabolism remains unclear. This study assessed the effects of exercise and exercise intensity on CD8+ and CD8+ subset metabolism and function. It was hypothesized that greater metabolic activity and glycolytic bias would be demonstrated Post- versus Pre- or 2h-Post-exercise (ex), and greater changes would be observed in more-differentiated subsets and following heavy versus moderate exercise. After enrollment and V̇O2peak assessment, participants (N=18, 29.2+/-6.9 years, 33.8+/-3.93 mL/kg/min. V̇O2peak, 9 Female) completed two exercise visits involving 30-minutes of cycling at moderate (-10% VT1) or heavy (+10% VT1) intensity. Blood samples were obtained Pre-, Post-, and 2h Post-ex to assess CD8+ metabolism and function. Effects of exercise, subset, and intensity were analyzed by linear mixed models. Bivariate correlations assessed relationships between subset proportions and metabolic regulation, and between CD8+ metabolism and function. Data were screened for statistical assumptions prior to analyses. The alpha-level was set to 0.05. Time and subset effects were detected for metabolic activity (F(2, 472.946)=52.696, p<0.001; F(4, 471.942)=224.487, p<0.001) and glucose dependence (F(2, 468.519)=10.295, p<0.001; F(4, 468.050)=65.951, p<0.001). A time*subset interaction was observed for mitochondrial dependence (F(8, 450.464)=2.996, p=0.003). Lower metabolic activity and greater mitochondrial dependence were observed Post-ex (p<0.01). Lower glucose dependence was observed 2h-Post-ex (p<0.05). CD8+ subsets exhibited expected metabolic profiles and responded to exercise similarly. However, effector memory (EM) CD8+ exhibited greater mitochondrial dependence 2h-Post-ex than Pre-ex (p<0.05). No significant effects were observed for metabolic regulation, nor were significant intensity effects found. Post-ex metabolic activity negatively correlated with cell function for all (p<0.05) but naïve CD8+ (p=0.238). EM demonstrated relationships between change in metabolic and cell function (p<0.05). In summary, acute aerobic exercise transiently decreased CD8+ metabolic activity and increased mitochondrial dependence, with increased metabolic flexibility 2h-Post-ex. Lower metabolic activity Post-ex was related to greater stimulated cell function, and change in EM metabolism related to change in function
Simultaneous State and Parameter Estimation Methods Based on Kalman Filters and Luenberger Observers: A Tutorial &amp; Review
No full textSimultaneous state and parameter estimation is essential for control system design and dynamic modeling of physical systems. This capability provides critical real-time insight into system behavior, supports the discovery of underlying mechanisms, and facilitates adaptive control strategies. Surveyed in this review paper are two classes of state and parameter estimation methods: Kalman Filters and Luenberger Observers. The Kalman Filter framework, including its major variants such as the Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), Cubature Kalman Filter (CKF), and Ensemble Kalman Filter (EnKF), has been widely applied for joint and dual estimation in linear and nonlinear systems under uncertainty. In parallel, Luenberger observers, typically used in deterministic settings, offer alternative approaches through high-gain, sliding mode, and adaptive observer structures. This review focuses on the theoretical foundations, algorithmic developments, and application domains of these methods and provides a comparative analysis of their advantages, limitations, and practical relevance across diverse engineering scenarios
Cognitive Processing with Emotion Interference in Bilinguals and Monolinguals
No full textResearch suggests that emotion is represented differently in bilinguals’ lexicon compared to monolinguals. Specifically, there are L1/L2 emotionality differences, such that messages are often judged as more emotional in L1, while L2 has been associated with reduced emotionality. Understanding emotion processing in bilinguals is important, as it may influence other areas such as decision-making, social interaction, and cognitive performance. The current project explored cognitive processing with emotional interference in bilingual adolescents and young adults of varying language backgrounds through secondary data analyses and newly designed experiments. The study included three components: (I) exploring emotional interference in adolescents using the emotional word-emotional face Stroop and emotional n-back tasks from the ABCD study; (II) examining the construct and external validity of emotional face stimuli, with a focus on how cultural background influences perception of facial emotions; and (III) testing young adults on six different tasks (variants of emotional lexical decision, Stroop, and n-back) to explore task-specific and language-related differences in emotion-cognition interactions. Across studies, we found that emotional and cognitive effects varied by task type and participant group, with language proficiency and age of acquisition influencing bilinguals’ performance. These findings help explain inconsistent results in the literature and highlight the importance of considering both task demands and individual language background. Overall, this work provides insight into how emotion, cognition, and language interact in bilinguals and lays the groundwork for future research using neuroimaging to explore the neural basis of these processes
Essays on Firms in International Trade
No full textThis dissertation has three chapters, each exploring a distinct dimension of how international trade affects firm behavior and labor market outcomes. In the first chapter, I argue that export participation provides incentives for investing in firm-provided training, thereby boosting human capital accumulation within firms. Using firm-level data for over 100 countries, I document that exporters are more likely to train their workers than non-exporters. Currency appreciations lead to more entry into exporting and an increase in the share of firms providing training, especially when they expose domestic producers to heightened import competition. These patterns are consistent with a heterogeneous-firm model in which exporting and training are complementary because of the higher revenues from exports and the productivity increase following training investments. By reallocating sales towards the most productive producers, import competition leads to a larger share of firms finding it profitable to export and provide training. In the second chapter, I focus on a specific channel for the link between exporting and training: quality upgrading. In particular, I show empirically and theoretically that, by encouraging quality upgrading, trade changes the incentives of firms to provide training to their workers. Using detailed firm-level data from Mexico spanning the period 1991-2004, I document that changes in the profitability of exporting lead to increases in the share of firms engaging in exporting, providing training to their workforce, and holding the ISO-9000 certification -my proxy for quality. Notably, these effects are stronger among larger firms. To establish causality, I employ two distinct empirical strategies: exploiting the heterogeneity in the impact of the 1994 Mexican Peso crisis on firms of varying sizes and utilizing variation in bilateral exchange rates interacted with predetermined industry-level trade shares. To provide a theoretical explanation for these findings, I develop a heterogeneous-firms model of trade that incorporates non-homothetic preferences, quality upgrading, and firm-provided training. In the third and last chapter, I study the distributional consequences of trade in economies characterized by distortions. Using aggregate data from over 100 countries, I find that trade liberalizations increase income inequality, particularly in less-distorted countries. To explain these patterns, I develop a trade model incorporating heterogeneous firms, skill-biased productivity, distortion wedges, and sorting of workers. Calibrating the model to microdata from firms and workers and to the Chilean trade liberalization of the 2000s, I show that distortions reduce gains from trade and mitigate increases in wage inequality. These results are consistent with firm-level empirical evidence from the same liberalization episode in Chile. My findings highlight a potential trade-off between welfare and inequality in second-best environments, providing new insights into the role of distortions
Task-Free Brain Breaks: The Perceived Impact on Student Behavior if Implemented in a Middle School Classroom
No full textBackground: Disruptive behaviors in middle school classrooms provide a significant challenge for teachers. Educators who face repeated inappropriate behaviors require new strategies to serve middle school students' needs that aim to reduce disruptive behavior (Emmer & Gerwels, 2013). While research exists regarding brain breaks with varying definitions, limited research exists regarding task-free brain breaks and their potential impacts on student behavior. Purpose: This qualitative study is intended to inform decision-makers and stakeholders how middle school teachers perceive the impact task-free brain breaks may have on student behavior if implemented in the classroom. Method: Through virtual, semi-structured one-on-one interviews, ten participants selected by their respective principal responded orally to questions guided by the three research questions. The overarching theme of the research questions addressed teachers' perceptions regarding the impact of task-free brain breaks on student behavior, the impact on student behavior over an extended period, and the fidelity of implementing task-free brain breaks. Furthermore, probing questions added and corroborated to the research in the literature review. Results: Data were analyzed through the Modified Van Kaam Method of Analysis of Qualitative Data. The review of participant responses indicated that while task-free brain breaks seem to be a strategy that may improve student behavior both immediately and over an extended period, fidelity was a concern and would be difficult to achieve without systematic implementation. Conclusion: Ultimately, teachers perceived that task-free brain breaks would have a positive effect on student behavior; they also felt other areas would be impacted, such as a reduction in stress, improved academics, increased instructional time, and reduced teacher stress. Further research to determine the effect of task-free brain breaks on student success would be beneficial. One recommendation for future research would be a mixed-methods study evaluating teachers utilizing task-free brain breaks in middle school classrooms
Crp/cAMP-Regulated Metabolism and Lipid Recycling in Stationary-Phase Persisters
No full textPersisters are a small subpopulation of genetically identical bacteria that can temporarily survive high concentrations of antibiotics without acquiring resistance. Once treatment ends, they can repopulate and cause infection relapse, eventually contributing to the emergence of resistant mutants. This thesis investigates persisters under nutrient-limited, stationary-phase conditions, which reflect natural and clinical environments where antibiotic tolerance is elevated. These conditions alter bacterial metabolism in ways that may support persister survival. In Chapter 1, I review current knowledge of persister metabolism, focusing on how energy production is maintained under stress. Although persisters are often described as metabolically dormant, emerging evidence suggests they engage in selective metabolic processes that are critical for survival. Chapter 2 builds on this background where I investigate cyclic adenosine monophosphate (cAMP) and its receptor protein Crp, which together form a global regulatory network that shifts metabolic activity from biosynthesis toward oxidative phosphorylation during stationary phase. Although persisters exhibit a reduced metabolic rate compared to actively growing cells, they remain dependent on energy metabolism. Multi-omic approaches including metabolomics, proteomics, and gene deletions consistently highlight the central role of the tricarboxylic acid (TCA) cycle, electron transport chain, and ATP synthase in sustaining persistence, all regulated by the Crp/cAMP complex. In Chapter 3, I examine the intracellular carbon sources that sustain these energy-generating pathways. Under nutrient-depleted conditions, lipid-derived glycerol emerges as a key substrate that fuels central metabolism and helps maintain the proton motive force. Disrupting this pathway by deleting genes such as tpiA and gloA, or by interfering with the TCA cycle, impairs energy homeostasis and reduces persister levels. These findings challenge the long-standing view of persisters as metabolically dormant and instead reveal a selectively rewired metabolic state maintained by internal carbon flux through phospholipid recycling. The final section outlines future directions to further define the metabolic strategies that support persistence. Using computational modeling and metabolic network analysis, we will investigate how these pathways function in more complex settings, including biofilms and in vivo infection models. Together, these studies provide a mechanistic framework for understanding energy-dependent survival in persister cells and suggest new approaches for targeting chronic bacterial infections