140816 research outputs found
Sort by
Strategies and Structures for the Honors First-year Seminar
No full textIntroduction to Honoring the First-year Seminar: Exploring High-impact Learning Experiences for the First Year in Honors (Lincoln, Nebraska: National Collegiate Honors Council, 2025)
Defining the Causal Contribution of the Gut Microbiota to Cardiomyopathy Pathogenesis as a Basis for Developing Microbiome-targeted Therapies
No full textHeart failure (HF) is a leading cause of morbidity and mortality worldwide, placing substantial clinical and economic burdens on society. Although advances in therapy have improved patient outcomes, the biological mechanisms that drive HF onset and progression remain incompletely understood. Increasing evidence suggests that the gut microbiota, the trillions of microorganisms residing in the gastrointestinal tract, plays a key role in some forms of heart disease, especially cardiovascular disease. Moreover, gut microbial dysbiosis has been associated with systemic inflammation, metabolic imbalance, and immune dysregulation, all of which contribute to the pathophysiology of HF. Cardiomyopathy, a primary disorder of the heart muscle, is a major precursor to HF and provides a tractable model for investigating how gut microbial communities may influence cardiac remodeling and functional decline. Yet, despite a growing recognition of the gut-heart axis, the causal contribution of the microbiota to the pathogenesis of cardiomyopathy remains poorly defined. The goal of this dissertation was to elucidate whether the gut microbiota contributes to cardiomyopathy and to identify potential avenues for therapeutic intervention. Our findings demonstrate that (1) alterations in gut microbiota composition were associated with disease severity during the progression of cardiomyopathy; (2) employing microbiome-targeted interventions that alter gut ecology influenced cardiomyopathy outcomes; and (3) the abundance of specific members of the gut microbiota correlate strongly with indicators of cardiomyopathy disease severity. By defining the microbial contributions to the pathogenesis of cardiomyopathy, this dissertation advances our understanding of the gut-heart connection and establishes a conceptual framework for microbiome-based strategies to prevent or mitigate HF. Collectively, these insights highlight the translational potential of targeting the gut microbiota as a novel avenue for maintaining cardiac health and managing chronic heart disease.
Advisor: Amanda Ramer-Tai
The Urban Heat Island and its Role in Shaping Convective Environments: An Observational Study
No full textThe continued spread of urbanization and its well-documented effects on regional weather and climate necessitate further research. Urban environments are often warmer and drier than their rural surroundings. This trade-off between a warmer but drier environment could have a complex impact on deep convection and its initiation. Previous studies have noted significant changes in precipitation patterns around cities as well as changes to ongoing deep convection. Given the well-documented theory and past research, it is hypothesized here that urban areas have a significant impact on the PBL thermodynamics of convective environments. The urban heat island (UHI) component of the Micro - Small-UAS Coordination for Atmospheric Low-Level Environmental Sampling (MicroSCALES-UHI) field campaign collected in-situ observations in and around Tulsa, Oklahoma, 9-13 September 2024. MicroSCALES-UHI used a network of uncrewed aircraft systems (UAS) to produce an in-situ dataset with high spatiotemporal resolution within an urban environment.
The goal of this work is to use this unique dataset to characterize the 4D impacts of urbanization on convective environments. Typical UHI modifications are observed with slightly warmer temperatures and lower mixing ratios than the rural surroundings. Analysis revealed that the urban modifications created areas of decreased total potential energy and moist static energy. This suggests decreased potential for deep convection initiation (DCI). In contrast, analysis revealed higher LCLs, suggesting possible increases to initial parcel/updraft size and an increased potential for DCI. Furthermore, steeper near-surface superadiabatic lapse rates, potentially enhancing buoyancy and lift, and a delayed transition from a convective boundary layer to a nocturnal boundary layer have implications for DCI and existing convection. Local zones of decreased buoyancy around urban centers suggest that while thermal perturbations from the UHI may influence DCI, other competing processes make the overall effect on DCI ambiguous.
Advisor: Adam L. Housto
On Bayesian Empirical Likelihood-based Method for Complex Survey Data with Application to Non-probability Sampling
No full textThis thesis develops a Bayesian empirical likelihood (BEL) framework for inference under complex survey designs and extends it to non-probability sampling. Parametric likelihood based methods are difficult to apply to complex survey data because the likelihood is rarely available in closed form. EL provides a flexible alternative by replacing the parametric likelihood with an empirical likelihood constructed from moment conditions. The proposed method first integrates empirical likelihood constraints with survey design features then extends BEL to non-probability sampling through selection models and design consistent restrictions. Posterior inference is carried out using a Metropolis–Hastings MCMC algorithm. A real-data analysis further illustrates the framework’s ability to correct selection bias when combining probability and non-probability samples.
Advisor: Sanjay Chaudhur
Reinforcement Learning Based Security Schemes for Distributed AI Systems
No full textDistributed machine learning (DML) is a component of modern intelligent systems, enabling collaborative training across devices such as mobile clients, vehicles, and edge networks. However, the decentralized nature of these systems introduces vulnerabilities, particularly data poisoning attacks that compromise model integrity and degrade performance. Traditional defenses, such as statistical filtering, robust aggregation, and privacy-preserving techniques, often struggle to adapt to overwhelming adversaries or operate under strict privacy and real-time constraints. This dissertation proposes the use of reinforcement learning (RL) and deep reinforcement learning (DRL) based misbehavior detection schemes that dynamically identify poisoning attempts in distributed AI systems, including federated learning, model parallelism and vehicular communication networks. The presented methods leverage behavioral patterns and aggregation-level indicators, such as gradient similarity, client contribution consistency, and clustering to learn optimal defense policies without requiring access to raw client data. First, an RL-driven trust management mechanism is developed to detect label-flipping and backdoor attacks in federated learning by tracking client updates over time and adaptively filtering malicious contributors. Second, a combined RL and graph-based supervised framework is introduced to detect falsified position reporting attacks in vehicular communication networks, improving resilience under dynamic mobility patterns. Finally, a scalable DRL architecture is designed to identify malicious clients in both federated and model-parallel training environments, demonstrating strong robustness even under severe adversarial conditions. Comprehensive experiments across MNIST, FMNIST, CIFAR-10, and real-world vehicular datasets show that the proposed solutions significantly improve detection accuracy, maintain global model performance, and generalize across attack intensities, surpassing existing benchmark defenses. Overall, this work advances adaptive security in distributed AI by establishing reinforcement learning as a powerful foundation for proactive, privacy-preserving, and scalable poisoning defense.
Advisor: Yi Qia
Making Deep Neural Networks Trustworthy: Intelligibility and Safety through Symbolic Methods
No full textThe rapid adoption of deep learning has come at the cost of properties long valued in artificial intelligence: intelligibility and safety. This dissertation develops methods that restore these properties by coupling neural networks with symbolic structure.
First, for supervised classification, I propose a differentiable decision tree integrated with a supervised variational autoencoder. The resulting model maintains competitive accuracy and generative performance while exposing clear macro-features in its latent space, improving interpretability.
Second, for reinforcement learning, I extend constrained Markov decision processes by specifying constraints in formal languages. This formal language constrained MDP enables the use of automata for state augmentation, cost shaping, and action shaping. This improves safety in exploration while remaining computationally tractable.
Third, I address safe exploration in nonlinear continuous-control tasks by integrating deep neural network learned dynamics with symbolic reachability analysis. Using star sets, I extract exact piecewise-affine dynamics and synthesize cycle-based trajectories that are provably safe with respect to the learned model. The hybrid architecture intervenes during reinforcement learning to prevent early failures, yielding both theoretical guarantees and empirical safety improvements in experiments.
Together, these contributions show that combining deep learning with symbolic reasoning can make machine learning models more intelligible and reliably safe, providing a foundation for trustworthy deployment in safety-critical domain.
Advisor: Stephen Scot
Real-time, Co-regulated Design for Cyber-physical, Multi-rotor UAS Swarms
No full textUncrewed Aerial Systems (UAS) have been integrated into a wide range of research and industrial applications, with growing interest in extending mission duration and spatial coverage through coordinated multi-UAS systems, or swarms. While swarming offers the potential for extended mission endurance and robustness through advanced path-planning, control, and estimation algorithms, significant challenges arise when implementing these methods on decentralized platforms composed of size, weight, and power-constrained (SWaP) vehicles. Limitations in onboard computational capacity and congested communication channels can break critical design-time assumptions, which at best, will degrade application quality of service, and at worst, destabilize the fleet through excessive delays and missed deadlines.
This research presents key breakthroughs in the design of SWaP-constrained UAS swarms for safe, efficient operation in outdoor, unpredictable environments. In this work, the limitations of SWaP-constrained vehicles are characterized through experimentation, and these observations motivate a co-regulated autonomy architecture for swarm control, navigation, and collision-avoidance that adaptively balances computational effort with the needs of the individual UAS and the swarm as a whole. Complementing this co-regulated architecture is a rate-adaptive real-time task model that further extends dynamic resource allocation for cyber-physical systems, ensuring that these adaptive behaviors may be safely implemented on constrained onboard hardware.
The resulting methodologies are validated through extensive experimentation on a UAS swarm platform, supported by a custom-built swarm toolchain for simulation, ground control, and communication. Data for this work was captured over the course of 300+ individual UAS flights that took place at Rodger\u27s Memorial Farm, NE and the Joint Inter-agency Field Experimentation (JIFX) at Camp Roberts, CA, with several flights of 8 UAS flying simultaneously. The experimental results show that state-of-the-art swarming capabilities can be made tractable on SWaP-constrained platforms by embedding resource awareness and flexible quality-of-service into the system design, enabling more efficient and more resilient cyber-physical autonomy.
Advisor: Justin Bradle
Integrating High Throughput Phenotyping Approaches to Characterize Plant Traits and Support Nutrient Management Research
No full textOver the past decade, considerable research has addressed the social, economic, and agronomic challenges posed by imbalanced nutrient applications, particularly nitrogen and, more recently, sulfur. In response, high throughput plant phenotyping, as one of the latest approaches, has contributed to the development of innovative techniques for assessing key plant phenotypes under stress over time in both field and greenhouse environments. However, there is a limited number of studies applying these techniques for nutrient management, particularly in scenarios where two abiotic stressors coexist, such as the combination of water stress and nitrogen deficiency, or concurrent nitrogen and sulfur deficiencies.
In chapter 2, high throughput plant phenotyping is leveraged to assess current nutrient management practices in rainfed and irrigated maize production for the field seasons of 2022 and 2023. Multimodal datasets from the NU-Spidercam Field Phenotyping facility site revealed a significant association between irrigation and improved crop productivity in the 2022 field season (ρ \u3c 0.05). Interestingly, in the 2023 field season, the observed relationships between irrigation and crop productivity were inconsistent due to environmental factors that critically influenced the establishment of the plant populations. This study’s innovative approach highlights the value of advanced remote sensing and plant phenotyping in enhancing trait evaluation for more efficient nitrogen utilization.
In chapter 3, high throughput plant phenotyping is used to investigate the interaction between nitrogen and sulfur levels in maize grown at the at the Greenhouse Innovation Center in Lincoln, Nebraska, (United States). Advanced imaging techniques were performed using visible and hyperspectral images to generate phenotypic time series and identify spectral features. Plant height and biomass accumulation revealed dynamic growth patterns, emphasizing a shift in resource allocation. Notably, integrating canopy location to extract the first derivative from hyperspectral images yielded initial insights for advancing nutrient-based stress detection (ρ \u3c 0.05), considering where chlorosis develops due to nutrient deficiency.
Collectively, these studies showcase novel applications of high throughput plant phenotyping to enhance nutrient management strategies in maize systems.
Adviser: Joe D. Luc
SoFi Technologies Incorporated Strategic Audit
No full textSoFi, a catchy abbreviation of Social Finance, is a digital bank as of 2022. The company seeks to offer better banking than traditional alternatives. SoFi, founded in 2011, has slowly developed financial offerings with the goal of dominating digital banking. The company is best known for helping customers manage student loans, but SoFi aims to provide customers with services for their every financial need. The platform boasts 10.9 million members, 33 billion in debt paid, and 117 billion in loans issued (SoFi). This report will explore SoFi’s origin, functions, and competitive environment. These investigations will inform recommendations for the company moving forward
Carrying Light in the Shadows: Four Portraits of Critical Literacy Teacher Agency in Unlikely Contexts
No full textThis qualitative study utilizes portraiture methodology to explore how educators in politically conservative and predominantly White schools enact critical literacy teacher agency. Using a theoretical lens of critical literacy and second-wave Whiteness studies, this research examines the lived experiences and instructional moves of four veteran English/language arts teachers. Through cross-case analysis, the study traces how teacher agency is contextually responsive and dependent on their professional identity, awareness of sociopolitical issues, and traditional school curricular expectations. The findings highlight how educators assess risk, navigate community norms, and maintain critical literacy through dialogic pedagogy and strategic adaptation. The four teachers in the study are professionals, carrying light in the shadows, as they take on the role of civic agents who preserve democratic ideals and protect intellectual freedom. This study broadens the field’s understanding of how socially conscious teaching endures, even under pressure.
Advisor: Loukia K. Sarrou