AUETD (Auburn University)
Not a member yet
9771 research outputs found
Sort by
Exploring LEO-Aided GPS Direct Position Estimation in Degraded Signal Environments
No full textDue to their low received signal power, Global Navigation Satellite Systems (GNSS) are easily subject to radio frequency interference (RFI). Subsequently, extensive research regarding advanced receiver designs that mitigate RFI is ubiquitous. Among these designs is the Direct Position Estimation (DPE) architecture, which addresses the shortcomings of conventional receivers by jointly processing all channels and estimating the receiver state in a single step. Combining each channel’s received power, this single-step methodology proves more robust than receivers that fuse measurements from independently processed channels in two steps. Despite this robustness, DPE can still succumb to the effects of RFI. This thesis discusses the performance capabilities of two DPE architectures that utilize dedicated and opportunistic low Earth orbit (LEO) positioning, navigation, and timing (PNT) sources, respectively, to supplement the Global Positioning System (GPS) in various RFI scenarios. Specifically, a Bayesian DPE approach is applied to each architecture, and necessary modifications are introduced for each LEO source. Furthermore, a methodology that prevents the obfuscation of GNSS information by high-powered LEO signals is presented.
Each architecture’s performance is evaluated using a Monte Carlo analysis that employs a correlator-level simulation of GPS L1 C/A. It is shown that including dedicated and opportunistic LEO sources substantially reduces the root mean square errors (RMSE) associated with the estimated position, velocity, and timing (PVT) states in high GPS attenuation regimes compared to other architectures. Specifically, the proposed architectures are compared to standalone GPS DPE, LEO-aided GPS Vector Processing (VP), and standalone GPS VP. The VP comparisons are included to gauge performance against a two-step methodology. The results also indicate that the probability of tracking GPS in the scenario with the highest RFI increases by up to 85.12 % compared to the additional architectures. Furthermore, a simple computational efficiency study assesses the benefits of aiding the naturally computationally expensive. DPE architecture with dedicated LEO signals. Lastly, an open-source satellite navigation simulation environment is introduced
Enhancing Hydrological and Climate Predictions Through Artificial Intelligence
No full textThe integration of artificial intelligence (AI) into hydrological and climate forecasting represents a significant advancement in enhancing prediction accuracy and reliability. This study explores how coupled land-atmosphere interactions influence soil moisture memory and water availability trends. In the Southeastern United States, the increased water use efficiency effect under elevated CO2 concentration is counteracted by the plant growth effect. To address the computational challenges posed by climate model experiments, specifically storage and runtime issues, this study developed two AI models:
- Hybrid Physics- AI model to improve hydrological forecasting.
-Deep learning model to improve ENSO prediction.
Building on insights into the interplay between biogeophysical variables, this research addresses the challenges presented by the National Oceanic and Atmospheric Administration’s (NOAA) high-resolution streamflow forecasting system using the National Water Model (NWM). We propose a hybrid Physics-AI model that integrates biophysical attributes such as topography, land use, and soil types with NWM forecasts, utilizing a deep learning approach to predict forecast errors. This hybrid model captures the complex interdependencies between biophysical attributes and hydrological processes, making it useful for predicting errors in areas lacking observational data.
Furthermore, integrating machine learning (ML) and deep learning (DL) models into ENSO index time series forecasting offers a promising approach to enhancing climate predictions. Traditional methods often struggle with accuracy and lead time, especially for long-term forecasts. However, recent advancements in DL models, such as Convolutional Neural Networks (CNNs), Bi-LSTM, hybrid CNN-1D and LSTM models, Gated Recurrent Units (GRUs), and Long Short-Term Memory (LSTM) networks, have demonstrated significant improvements in predictive skill. These advancements are influenced by lookback periods, where analyzing dependency on different historical data spans optimizes model accuracy. High-quality datasets further enhance model performance, leading to more reliable forecasts.
By comparing various DL models, we identified unique advantages, such as Bi-LSTM and LSTM networks' ability to capture temporal dependencies and GRUs' strength in feature extraction. Advanced training techniques like transfer learning bolster these models' efficacy. In conclusion, this research significantly enhances hydrological and climate predictions by integrating AI with traditional models and employing advanced ML/DL techniques for ENSO forecasting. The hybrid Physics-AI model improves forecast reliability from 21\% to 82\% compared to the NWM. For the 1-12 month lead period, skill transfer learning improved ENSO prediction skill, achieving an anomaly correlation metric (ACM) consistently above 0.8, comparable to Wang et al. For the 12-24 month lead period, the hybrid CNN-1D-LSTM model maintains an ACM above 0.5, though Wang et al.’s model shows superior long-term prediction skill with a higher and more stable ACM. The hybrid model, trained on the CESM2-LE-EOF dataset, excels in short-term forecasts, while Wang et al.’s model is more robust for long-term predictions. These complementary strengths will aid in better water resource management and climate risk mitigation
Harnessing visual context information to improve face identification accuracy and explainability
Get PDFFace identification (FI) is ubiquitous and drives many high-stake decisions made by the law enforcement. A common FI approach compares two images by taking the cosine similar- ity between their image embeddings. Yet, such approach suffers from poor out-of-distribution (OOD) generalization to new types of images (e.g., when a query face is masked, cropped or rotated) not included in the training set or the gallery. Recently, interpreatable deep metric learning with structural matching (e.g. DIML [101] and Vision Transformers [27]) obtained significant outcomes in popular computer vision problems such as image classification, image clustering, etc. In this proposal, we present simple yet efficient schemes to exploit structural similarity for an interpretable face matching algorithms. We propose two following novel methods.
• DeepFace-EMD: A re-ranking approach that compares two faces using the Earth Mover’s Distance on the deep, spatial features of image patches.
• Face-ViT: A novel architectural design using Vision Transformers (ViTs) for out- of-distribution (OOD) face identification and show significant improvement in inference speed. We feed embeddings of both images through a pre-trained CNN by ArcFace [22], layers of a Transformer encoder, and two linear layers as part of a ViT. We train the model with 2M pairs sampled from the CASIA Webface [93].
Our extra comparison stage explicitly examines image similarity at a fine-grained level (e.g., eyes to eyes) and is more robust to OOD perturbations and occlusions than traditional FI. Interestingly, without finetuning feature extractors, our method consistently improves the accuracy on all tested OOD queries: masked, cropped, rotated, and adversarial while obtain- ing similar results on in-distribution images. Moreover, our model demonstrates significant interoperability through the visualization of cross-attention
First-Generation OTC Antihistamine Use and Voice Function: A preliminary study
No full textObjectives/Hypothesis: The primary goal of this investigation was to characterize the effect of the first-generation, over-the-counter antihistamine, Chlor-Trimeton on laryngeal structure and function.
Study Design: Prospective within-participant multimodality repeated measures design.
Methods: Eight consented participants (seven females, one male) completed the study. Volunteers completed the following measures before and 2 hours after antihistamine: perceptual vocal function measures, phonation threshold pressure, acoustic measures, and laryngeal imaging. Paired t-tests were used to determine statistical significance of change pre- and post-administration of the antihistamine. Laryngeal imaging data were descriptively analyzed.
Results: A positive correlation between the OMNI-Vocal Effort Scale and the Rate of Fatigue index was identified. No other significant differences were identified for any measures taken. Descriptively, all participants had evidence of mucosal changes in the form of one or more of the following: increased vascularity, mucus in the anterior commissure, and vocal fold color changes (i.e., white to red), all of which are consistent with prior descriptions of allergy larynx.
Conclusions: Empirical study of laryngeal appearance and function changes post administration of a commonly used OTC antihistamine affirmed clinical observations of laryngeal tissue changes that have been considered typical for individuals with upper airway allergies. Further study of this population with and without diagnosis of voice impairment is warranted
Application of In-Silico Protein Engineering and Optimization Methods to Design Target-Specific Antibody Mimetics
No full textAntibodies (Abs) are proteins that bind target antigens (Ags) and interact with immune receptors. Over the years they have been developed into therapeutics with tremendous potential to treat various diseases. Their global market share is projected to cross $500 Billion by 2030. As excellent as therapeutic antibodies are when administered to patients, they exhibit limitations associated with solubility, aggregation propensity, and degradation tendency when working outside the body. Therefore, it is important to move beyond conventional Abs and develop alternative protein binders that can mimic their functions while being able to address these limitations at the same time.
Previously, several works have developed in-silico algorithms to design functioning proteins to bind their targets. However, they typically generate many predictions, which are practically unfeasible for experimental testing due to high cost and time constraints. This dissertation demonstrates the application of complex scientific protocols to shortlist the most promising computational designs from a large dataset of predictions most worthy of experimental testing.
Chapter Two will use a novel PETEI algorithm to design thousands of 10th fibronectin type III domains to bind 3 distinct tag epitope peptides, FLAG, HA, and MYC. Since it is unfeasible to test all the designs, they will be energy minimized using CHARMM and Rosetta forcefields based on three or fewer positive CDR residues. The Rosetta Interface Analyzer will calculate and analyze their binding energy per buried surface area. Similar binding metrics will be calculated for an energy minimized, non-redundant 231 Ab-protein database and compared to the designs. The designs will be rank ordered and the top 30 will be selected for experiments. This chapter is part of a bigger project, submitted for publication as of July 2024.
3
Chapter Three will use another novel algorithm called MutDock to design 2145 DARPin – α-Cobratoxin complexes. They will be down-selected to the top 39 using the scientific protocols mentioned previously, and affinity matured using specific RosettaDesign protocols to improve binding. 3900 affinity matured designs will be narrowed down to 73 by applying similar protocols and visually inspected in UCSF Chimera. Short 5 ns NAMD MD simulations will be conducted to assess their structural integrity over time and conformational binding metric trajectory will be plotted against time to analyze their characteristics. Their standard deviations will be calculated and this analysis will assist in down-selecting the top 51 designs for experiments.
Chapter Four will use tools like CamSol, HoTMuSiC, SCooP, IEDB De-immunization, and RosettaDesign to fine-tune the physicochemical properties of 5 α-helical bundles to arrive at the desired protein. These top 5 were shortlisted from a list of 75 similar structures after discarding the ones with Cys residues, de novo designs, extra end loop, and the position of an end loop to cause steric hindrance. RosettaFold predicted 1 out of 5 re-designed sequences to fold to their native state. This binder will be the starting point for the next 2 more iterative rounds of fine-tuning to generate 16 different ‘ideal’ binding proteins. Therefore, observably, the level of complexity of the development and applied scientific protocols implemented in the subsequent chapters keeps increasing progressively which leads to a more sophisticated down-selection process for designs worthy of experiments.
Chapter Five will elucidate the usage of I-Tasser, Robetta, and trRosetta to model 3 anti-CTLA4 nanobodies and canine CTLA4 protein. ZDOCK was used to dock them, which resulted in 85 of the 90 predictions showing the nanobodies binding to the CTLA4 conserved epitope. This was a modeling of experimental results, and it was published in Scientific Reports. Chapter Six will explain the future directions of the current projects
Alabama Early Childhood Education Leadership: COVID-19, Where Did the Impact Leave Us?
Get PDFThis study explores the experiences of ten leaders within the Alabama Department of Early Childhood Education (ADECE) during the COVID-19 pandemic. As the pandemic forced closures and adaptations across early childhood education programs, these leaders played a crucial role in maintaining operations and supporting stakeholders. The research examines how they navigated the challenges of closures, reopenings, and evolving safety protocols while ensuring the well-being of staff, students, and teachers. Through interviews with the ADECE leaders, the study identifies recurring themes that shaped their experiences. It explores how the ADECE adapted its policies and procedures while ensuring core principles remained intact. The research and this study highlight the leadership approaches that fostered sustainability within the ECE system.
Additionally, the research examined three leadership theories relevant to the needs of Early Childhood Education (ECE) organizations during crises: distributive leadership theory, talent-centered leadership theory, and sensemaking theory. The findings underscore the importance of adaptable leadership styles, such as distributive leadership, in crisis situations. This study offers valuable insights for current and future leaders in the Alabama Department of Early Childhood Education (ADECE), providing strategies for navigating uncertainty and fostering a more resilient ECE system. This research serves as a reference point for the ADECE's ongoing development and preparedness for future challenges
Genetic Biotechnology to Improve Reproduction of North American Catfish for Aquaculture, Genetic Enhancement and Genetic Conservation
Get PDFXenogenesis is an emerging technology for hybrid catfish production using primordial germ cells (PGCs), spermatogonial stem cells (SSCs), or oogonia stem cells (OSCs) transplanted to a sterile host species. The present study investigated the recovery of spermatogonial stem cells through short-term culture before transplantation using various incubation conditions. Stem cell research is a rapidly growing area that has the potential to generate therapeutic drugs to treat diseases as well as study disease progression from the beginning for humans. Many of the same techniques apply to human pluripotent stem cell culture as they do to normal mammalian cell culture. However, maintaining the undifferentiated state of human pluripotent stem cells (hPSCs) requires extra considerations to ensure that the cells keep their key traits of self-renewal and pluripotency. Such information is not available for blue catfish, Ictalurus furcatus, to enable catfish xenogenesis research. SSCs were extracted and isolated from the immature gonads of blue catfish. The maintenance of spermatogonial stem cells was investigated in this work used short-term culture prior to transplantation. The SSCs were incubated with and without 5% CO2 at 24-30°C temperature range and with 0-75mM or ROCK I. Spermatogonia produced in vitro discovered to be the best treatment at 30°C, with 50-75 mM ROCK I with 5% CO2 (p=0.001) for 72 hours.
The timing of implantation and the number of transplanted cells were varied in order to optimize the selection of channel catfish (Ictalurus punctatus) fry and to evaluate the proliferation of donor cells into functional gonadal tissues. Xenogenic progeny were successfully produced in channel catfish when donor cells were implanted between 3 and 5 days post-hatch. Further experiments aimed to assess the efficiency of germ cell transplantation by introducing spermatogonial germ cells from blue catfish and channel catfish into sterile common carp fry, resulting in xenogenic common carp capable of producing channel or blue catfish gametes. The most effective injection window was determined to be between 15 and 25 days post-hatch.
The effects of hormonal treatments on spawning success and reproductive performance in CRISPR/Cas9-generated melanocortin-4 receptor (mc4r) knockout channel catfish (Ictalurus punctatus) were investigated. Varying hormonal regimens, including luteinizing hormone-releasing hormone analog (LHRHa) and human chorionic gonadotropin (HCG), were evaluated for their impact on spawning rates, relative fecundity, hatch rates, and fry yield per kilogram of female body weight. Results demonstrated that HCG was crucial for successful spawning in mc4r mutants, with spawning failure observed in its absence, despite the presence of pronounced secondary sexual traits. The combination of HCG and LHRHa significantly enhanced reproductive outcomes, with mc4r x mc4r, pairings exhibiting fecundity and hatch rates comparable to wild-type controls under optimized hormonal protocols
Vietoris-Rips and Čech Complexes of Certain Finite Metric Spaces
No full textWe examine the homotopy types of Vietoris-Rips complexes on different collections of subsets of equipped with the symmetric difference metric. More specifically, we prove that the Vietoris-Rips complexes \V(\F;2) and \V(\F\cup\FF;2) are either contractible or homotopy equivalent to a wedge sum of 's. We also show that the complexes \V(\F\cup \FFF;2) and \V(\F_{\preceq A};2) are homotopy equivalent to a wedge sum of 's. We found a more intuitive proof of the result of Adamamszek and Adams in \cite{AA22} about Vietoris-Rips complexes of hypercube graphs with scale .
We also define \v{C}ech complexes on the finite union of finite metric spaces at scales equipped with symmetric difference metric. We prove that the \v{C}ech complexes\\
\N(\F\cup \FF\cup \FFF;2) and \N(\F\cup \FF\cup \FFF;3) are either contractible or homotopy equivalent to a wedge sum of 's and that the complex \N(\F\cup \FF;2) is homotopy equivalent to a wedge sum of 's. We also establish that the complex \N(\f_0^m\cup \f_1^m\cup \f_2^m\cup \f_3^m;3) for is homotopy equivalent to a wedge sum of 's. We provide (inductive) formulae for all these homotopy types
1H Nuclear Magnetic Resonance Investigation of Pain Processing in the Human Somatosensory Cortex
No full textThe perception of pain is an adaptive trait critical to survival. Prolonged painful stimulation in the form of chronic pain, however, can be maladaptive and cause significant physical and emotional distress. Indeed, chronic pain is linked to substantial health and economic burdens. To improve pain-related outcomes and to advance the field’s understanding of the neurochemical mechanisms of normative pain processing, the current study used functional magnetic resonance spectroscopy (fMRS) to investigate glutamate and γ-Aminobutyric acid (GABA) levels in the primary somatosensory cortex (SI) in response to acute, pressure-based pain. Additionally, the current study investigated the link between pain physiology and perception by assessing the relationship between pain-related neurometabolites and subjective pain ratings. Data suggest that pain was not associated with changes in either glutamate or GABA. Further, linear regressions did not reveal any predictive relationship between glutamate or GABA and subjective pain ratings. Finally, statistical analysis revealed a significant reduction in total creatine levels pre- versus post-task. These data add to the growing corpus of spectroscopic pain literature and to the gap in current literature regarding the role of SI in normative pain processing
Captioning of Non-Speech Information in Children’s Programming on Pbskids.org
No full textThe availability of closed captions in children’s programming has increased in recent years. However, quality, consistency, and other aspects of closed captions is still lacking, especially with non-speech information (NSI). The Described and Captioned Media Program (DCMP) provides a Captioning Key to assist with these aspects, yet there is little research looking at how well current closed captions follow the suggestions within the Captioning Key. This study provides a more in-depth look at closed captioning practices related to NSI among currently produced children’s programming distributed through pbskids.org. As PBS KIDS is a DCMP Partner and has shown to be a leader in captioning, it should be an exemplar in this area. The results give benchmarks for future research in captioned NSI and show that captions in current programming on pbskids.org typically contain a high prevalence of captioned NSI, with Speaker IDs being most prevalent among the majority of episodes. Also, while consistency of captioned NSI is high within each episode, it is not across the sample, and compliance with DCMP’s Captioning Key is minimal. This points to the need for more standardization among captions in children’s programming distributed by PBS KIDS which could be improved with the adoption of the Captioning Key or a variation tailored toward children’s programming