SHAREOK Repository
Not a member yet
49261 research outputs found
Sort by
ENTROPY INTEGRATED DYNAMIC ROUTING IN CAPSULE NETWORKS AND DEEPFAKE DETECTION APPLICATIONS
No full textCapsule networks are recognized for their ability to capture part-whole hierarchies in visual data, but they face challenges in managing uncertainty within part-object relationships. This study introduces an entropy-integrated dynamic routing algorithm to enhance both the performance and interpretability of capsule networks by embedding information-theoretic principles. Our method incorporates an entropy-based regularization term in the final iteration of the routing process, improving routing decisions and reducing reliance on uncertain capsule connections. Evaluations on CIFAR-10 demonstrate a mean accuracy of 85.58\%, surpassing the baseline accuracy of 84.61\% achieved with standard dynamic routing. Comparative analysis with recent entropy-based routing approaches highlights our method’s balance of computational efficiency and routing flexibility, achieved without additional model complexity. These findings position entropy-integrated dynamic routing as a powerful tool for enhancing the interpretability and effectiveness of capsule networks in high-uncertainty environments. Our study finds application in deepfake detection by embedding entropy into the capsule networks’ dynamic routing process. Traditional capsule networks capture spatial hierarchies critical for identifying synthetic media, yet their scalability is limited by uncertainties in iterative routing. To address this, we put into practice our entropy-based regularization that adjusts routing coefficients based on capsule activation uncertainty, selectively prioritizing reliable connections and enhancing interpretability and model robustness without significant added complexity. Experiments on FaceForensics++ and DeepPhy datasets reveal that our entropy-integrated routing algorithm achieves higher classification accuracy in both binary and multi-class deepfake detection tasks, outperforming baseline models like CapsuleForensics. These results underscore entropy regularization’s potential to improve capsule networks’ capacity to address nuanced spatial inconsistencies in manipulated media, advancing scalable and interpretable solutions in deepfake detection
INVESTIGATION OF APERTURE LEVEL SPATIAL AND FREQUENCY FILTERING ON RADAR AND COMMUNICATION SYSTEM PERFORMANCE
No full textDynamic spectrum access (DSA) and cognitive radio (CR) systems, built on software-defined radio (SDR) platforms, enable adaptive spectrum use and reconfiguration. Reconfigurable and filtering antennas play a crucial role in DSA, reducing the need for extra filtering circuits and rejecting out-of-band (OOB) signals. However, a comprehensive analysis of their role in SDR-based DSA system performance is yet to be conducted. This work introduces a novel SDR-based test system emulating frequency-variant interference, evaluating traditional broadband, reconfigurable, and filtering antennas by measuring received interference power and signal-to-interference plus noise ratio (SINR) in congested environments. The study shows that reconfigurable filtering antennas enhance OOB interference rejection and offer superior performance in size, weight, and power (SWaP) constrained environments, highlighting their potential for improved system-level performance in congested spectrum scenarios
INVESTIGATION OF DIMENSIONALITY REDUCTION TECHNIQUES IN THE SURROGATE MODEL DEVELOPMENT FOR SPATIOTEMPORAL STORM SURGE PREDICTIONS
Get PDFCoastal communities face unique challenges due to their increased vulnerability and exposure to natural hazards, with humans and their assets being at a close proximity to the ocean and to a variety of severe weather phenomena such as tropical cyclones. Storm surge, which is the abnormal rise in seawater level due to storm winds, can lead to catastrophic flooding, erosion and infrastructure damage, putting lives and properties at risk. With the frequency of extreme weather phenomena like tropical cyclones increasing due to climate change the need to be able to predict their impact in coastal areas is increasing. Accurate predictions of storm surge, as a tropical cyclone is approaching, are crucial for informing evacuation planning efforts, allowing communities to have enough lead time to prepare and respond effectively. Through robust predictions which provides timely, location-specific information, such vulnerable regions resilience can be improved by saving lives and reducing property damage, thereby aiding recovery efforts. In this effort to obtain reliable predictions, advances in computational and numerical models, have enabled the development of physics-based models, that can predict storm surge based on complex fluid dynamics equations and atmospheric conditions. These models are the closest approximation possible to the physical phenomena that contribute to the experienced surge, providing valuable and detailed insights for large regions of interest. Unfortunately, due to their inherent complexity, they are computationally intensive, requiring large memory and number of processors, with the time for each analysis necessitating a couple of days in a state-of-the-art supercomputing facility. These limitations prevent the use of such accurate and detailed models during real-time risk assessment efforts, where a specific tropical cyclone has been formed and is hours away from landfall. Especially these settings, it is of crucial importance to be able to produce time-series storm surge estimates of high spatial resolution fast and accurately so that preparedness and recovery efforts can be guided appropriately and utilized efficiently, maximizing their impact to the extent possible. In an effort to overcome these limitations, surrogate models have emerged as a versatile alternative, that offer a fairly simple, tractable, mathematical approximation of a computationally expensive model, allowing the production of fast and accurate surge predictions, provided that they are properly calibrated on available datasets for the region of interest. Such efficiency without the significant loss of any accuracy makes them a promising alternative in real-time emergency planning and response when resources and time to respond are limited. This requires the development of surrogate models that will be able to handle large geospatial regions providing detailed predictions in the most computationally efficient way to allow for enough lead time for the evacuation managers to develop their plans, exploring at the same time a large number of potential stormevolving scenarios Pursuing model simplicity and computational efficiency, past efforts have explored the development of surrogate models of different mathematical complexity, that leverage the underlying spatial correlation that the grid locations have within the domain of interest. This allowed for linear and nonlinear dimensionality reduction techniques like Principal Component Analysis (PCA) and AutoEncoder (AE) architectures to generate latent space representations, that can be used as the reduced size input during the surrogate model development and deployment phases. After predictions are made in that reduced, latent space, the actual surge predictions are compiled through an inverse mapping. So far, such investigations have not thoroughly examined nonlinear dimensionality reduction techniques for large geospatial domains where time-series predictions are warranted, especially for locations that are onshore where highly nonlinear surge behavior may be recorded during different tropical cyclone events. This thesis advances these efforts by performing a detailed investigation on the development of a nonlinear inverse mapping (utilizing the versatile family of autoencoders) designed to capture non-linear relationships for onshore nodes tackling challenges related to providing time-series storm surge predictions over a large number of regions. The latent space will be used as input in a Gaussian Process surrogate model to provide storm surge time-series predictions on a storm suite that the models have not been trained upon. This development seeks to support high-resolution forecasting, making it suitable for real-time emergency response as well as for long-term planning and infrastructure design for any region of interest, given the existence of a well-calibrated region-specific surrogate model
Mass spectrometry-based metabolomics analysis of Chagas disease models
No full textChagas disease (CD) is a neglected tropical disease which is caused by the parasite Trypanosoma cruzi. Approximately 6 to 7 million people in the world are infected with T. cruzi. The symptoms at the acute phase are mild or non-specific, so most people won't get treatment, which leads to the chronic phase. 20% to 30% of people in the chronic phase will develop Chronic Chagas Cardiomyopathy which is the primary reason for death in CD cases. Current drug treatments are limited and have to be given in the acute phase for best efficacy. Therefore, the major challenge of CD is prevention for people in endemic regions and treatment during the chronic phase. This dissertation applied untargeted and targeted metabolomics based on mass spectrometry to investigate the metabolic perturbation induced by progression to symptomatic disease after infection, treatment effects of vaccine-linked chemotherapy and protection effects of prophylactic vaccines. We combined LC-MS/MS with various data science methods for higher resolution and deeper insights in complex biological samples. Our findings extend knowledge about disease progression and failure of traditional treatments. In addition, we evaluated novel vaccine candidates for better protection and treatment efficacy. Clearance of the parasite does not restore metabolites, but successful restoration of metabolites by a therapeutic vaccine combined with benznidazole led to better health status in chronic mouse models. Immune responses from single candidate vaccines showed better protection than combined immunization. Multiple metabolites, including lipids, amino acids and nucleotides showed strong correlation with multiple immune response transcriptome and miRNA expression indicators. The NGP24h vaccine showed the largest difference for these metabolites compared to other vaccine groups. However, there were few metabolites correlated with the parasite load. This distinction between parasite burden and metabolic changes was found across the projects reported here. Overall, these results reveal molecular mechanisms of CD treatment and immune response protection, based on simultaneous effects on the pathogen and on host small molecule responses. This link between infection and subsequent persistent small molecule perturbation broadens our understanding of infectious disease and has potential applications for other disease models
INTEGRATING GENOMIC, MORPHOLOGICAL, AND PALEONTOLOGICAL DATA TO UNRAVEL MACROEVOLUTIONARY DRIVERS OF MORPHOLOGICAL DIVERSITY IN FISHES
Get PDFPerhaps one of the most apparent, yet captivating, natural phenomena is the diversity of shape and form which has evolved on Earth. Morphological evolution is shaped by numerous factors across multiple biological scales, yet there is still much to be discovered using novel techniques and integrative approaches. By combining insights from phylogenetic comparative methods, paleoclimatic models, geometric morphometrics, and comparative transcriptomics, my dissertation provides a multi-faceted approach to understanding the factors (ecological, environmental, and genetic) contributing to the evolution of body size and shape in two diverse groups of fishes.
In Chapter 1, I use paleoclimate data in conjunction with a newly inferred phylogeny based on both extant and fossil species to examine how past ocean temperature is correlated with body size in tetraodontiform fishes (pufferfishes, boxfishes, ocean sunfishes, and allies). Numerous rules exist, which attempt to summarize the evolution of body size. These include Cope’s rule, which states lineages tend to increase in size over evolutionary time scales, and Bergmann’s rule which posits species tend to be larger in colder environments and smaller in warmer environments. These rules are generally well-supported in endotherms, but remain poorly understood in ectotherms. Using tetraodontiform fishes as a model clade, owing to their robust fossil record and disparity in body sizes, I find strong support for increasing body size over time in relation to decreasing oceanic temperatures. These results highlight the impact of paleoclimatic changes on aquatic ectotherms, which depend on their environment for temperature regulation and are potentially more susceptible to climatic changes compared with terrestrial vertebrates.
In Chapter 2, I continue investigating morphological evolution of tetraodontiform fishes, a clade that is extremely well-suited for these types of questions due to their extraordinarily unique morphological diversity, including spines and spikes in porcupinefishes, box-like armor in boxfishes, and inflatable bodies in pufferfishes. Here, I utilize data from three-dimensional CT scans of both fossil and extant species to investigate widescale drivers of morphological evolution in relation to habitat and key innovations. Habitat transitions and evolutionary innovations have been previously linked to increases in morphological diversification, but it is unclear whether these are universal drivers. Using tetraodontiform fishes as a model system, I show that these general rules may be more nuanced than previously thought. Coral reefs have long been suggested to increase morphological diversification in fishes, however I find that species living in other habitats display higher rates of skull shape evolution, suggesting reef association alone is not sufficient to spur high evolutionary rates. Additionally, I investigate a morphological novelty—the tetraodontiform beak—which is a fusion of the teeth into a beaked mouth in several families. I find that beaked families exhibit higher rates of morphological evolution compared with non-beaked families, suggesting that the beak may be an evolutionary innovation facilitating their diversification.
Lastly, in Chapter 3, I investigate morphological evolution through a genetic lens by employing comparative transcriptomics and differential expression analyses to identify candidate genes involved in miniaturization in gobiid fishes. While large body size is traditionally seen as advantageous, numerous transitions to miniaturization, the extreme reduction of adult body size, have evolved across the Tree of Life. Despite how common miniaturization appears, its genetic mechanisms are poorly understood. Miniaturization is especially common among fishes, with species in the family Gobiidae (gobies) being an exceptional case. Gobiid fishes are part of an ecological grouping called “cryptobenthic reef fishes” which are the poster child for small-bodied fishes. Even within the already small-bodied gobiid phylogeny, there are multiple, independent transitions to extreme small size, allowing for tests of genetic convergence within a comparative macroevolutionary framework. Here, I assemble the first de novo transcriptomes for six species of gobiid fishes, which represent three clades each containing a closely related large-bodied and small-bodied species. I identify sets of statistically significant orthologs which are differentially expressed between large-bodied and small-bodied species in each clade. From these, I identify several candidate genes potentially involved in miniaturization, including ybx1 and bzw2, both known to affect cell growth and development. These candidate genes offer insight into the genetic convergence on miniature body size and provide a framework for future studies.
Overall, my dissertation provides new insights into the large-scale processes and dynamics which have shaped the evolution of morphological diversity in fishes. By combining data from both fossil and extant taxa, as well as analyzing the evolution of size and shape through both a morphological and genomic lens, we can appreciate the complexities and nuances of morphological evolution and gain a more complete picture of the evolutionary processes which shape life on our planet
Subgrid Scale Modeling of Turbulence and Cloud Microphysics Interactions
No full textClouds have a significant but uncertain impact on Earth's climatological and hydrological cycles. In particular, the warm rain process has considerable inconsistencies between current theories and observations. One hypothesis to explain this is the role of turbulence in broadening the droplet size distribution. The primary methods of studying these processes have been through laboratory studies, direct numerical simulations, and large eddy simulations. However, these processes are difficult to study as they occur on a broad range of scales. This makes large eddy simulations appealing as they remain computationally efficient by modeling the smallest, dissipative scales of motion. While past studies have made efforts to improve the subgrid-scale stress tensor and scalar flux vectors in large-eddy simulations, subgrid-scale terms related to cloud microphysics have received little attention. Specifically, subgrid-scale supersaturation variance is important when considering a Lagrangian microphysics approach as it arises from the Langevin equation, and both subgrid scale supersaturation and concentration covariance and subgrid scale concentration variance arise from the filtered evolution equation for droplet size distribution. It is these terms that were the focus of this study.
This study computed the true subgrid-scale variance and covariance terms from data of direct numerical simulations of Rayleigh-Bénard convection in the Michigan Technological University Pi Chamber. Five cases of varying aerosol injection rates were considered, each with a Rayleigh number of 7.9x10^6. The true subgrid-scale terms were compared to two candidate models: the gradient model and the scale-similarity model. Statistical analysis consisting of probability density functions, joint probability density functions, and correlation coefficients was used to assess model performance. Results concluded that the gradient model had relatively poor agreement with the true subgrid-scale terms with joint probability density functions that did not follow the one-to-one line which would indicate good skill, and correlation coefficients between 0 - 0.4. In contrast, results from the similarity model indicated joint probability density functions that closely followed the one-to-one line, and correlation coefficients between 0.3 - 0.9 suggesting good agreement between the true and modeled subgrid-scale term. Altogether, the similarity model showed promise for modeling the subgrid-scale supersaturation variance, supersaturation and concentration covariance, and concentration variance. However, future investigation with higher Rayleigh numbers is warranted where larger scale separation exists between the large and small scales of turbulence
Deeper Listening: Aural Learning as a Tool for Large Instrumental Ensemble Rehearsals
Get PDFMusic is primarily an aural experience, yet much of the pedagogy utilized in wind band rehearsals in the United States is reliant on the reading of notation and the use of words to explain musical ideas to an ensemble. While Western notation may serve as an efficient transmission method for rhythmic and pitch content, it does not as effectively convey other musical elements such as timbre, articulation, dynamics, style, and emotion. Music educators in the United States desire for students to develop listening skills, but does the teaching style dominated by reading visual notation afford students opportunities to fully develop their listening skills and, ultimately, musicality?
Music is a multi-sensory phenomenon, which implies that music pedagogy should engage the auditive, kinesthetic, and visual senses. Western art music pedagogy began as an aural experience before the rise of Western notation, and countless other musical cultures rely on aural learning as their primary transmission model. By adding aural learning practices to U.S. wind band rehearsals, students may develop a deeper and broader understanding than through reading notation alone, potentially fostering social and cultural understanding between students and other musical cultures. Through this research study, I argue that Irish traditional music pedagogy (which is based on aural learning) can be an effective teaching method in a wind band setting. I highlight this through an analysis of the histories of aural learning, the application of an aural learning model for large ensemble rehearsals derived from Irish traditional music practices, and an analysis of the perceived impact of aural learning pedagogy on the collegiate wind band students
EXPERIMENTAL AND THEORETICAL STUDY OF A NOVEL FREEZE DESALINATION METHOD USING AN INTERMEDIATE COOLING LIQUID
Get PDFFreeze desalination (FD) emerges as a promising method for treatment of high-salinity brines. In this work, experimental and theoretical studies are conducted on the design, fabrication, and testing of a novel FD system for desalinating brines with salinities up to 100,000 ppm. The system integrates a refrigeration unit with a desalination unit via an intermediate cooling liquid (ICL). The desalination unit is comprised of a freezing chamber, a slurry transport section, and separation modules. Operating at atmospheric pressure, the FD system leverages the efficient heat transfer achieved through direct contact between the brine and the ICL, thus avoiding complications like ice adhesion to cooling surfaces and the mixing of refrigerant with the treated water. A key factor in increasing the energy efficiency of the developed method is recovering the cold energy of the generated ice to cool the condenser of the refrigeration unit. This can be achieved by running an ice-water slurry through a heat exchanger to absorb heat from the condensing refrigerant. To better understand the fundamentals of ice-water slurry heat transfer, a computational model is developed to simulate the melting of a suspended solid particle in its own melt. The fabricated prototype is used to study the impacts of various operational parameters such as feed brine salinity, cooling temperatures, and centrifugation times on the recovery ratio and the purity of the treated water. It is found that lower cooling temperatures and feed brine salinities enhance the recovery ratio, while increasing the salinity of the treated water. Specifically, a feed brine with a TDS of 70,000 ppm and a cooling temperature of -17°C resulted in a recovery ratio of approximately 50% and a treated water TDS of about 2,600 ppm. This study demonstrates a novel method for desalinating high-salinity brines, which potentially offers greater energy efficiency compared to conventional evaporative methods. This method could have significant applications in industrial brine management and brine mining
Stressed Out: Resistance-nodulation-division (RND) and Major Facilitator Superfamily (MFS) Efflux Pump Mediated Survival Strategies of Acinetobacter baumannii
No full textOriginally discovered in soil microbial communities, reports of genus Acinetobacter associated infections started to increase around 1977 and continued to climb to frightening heights for the next 20 years resulting in Acinetobacter becoming a high priority target for antibiotic resistance research, leading to the research detailed here. A. baumannii’s exceptional abilities to evade antimicrobials have long been known, but the details of those mechanisms have for the most part remained unclear. The better we can understand the pathogen’s physiology, the more informed our efforts toward alternative treatments can be.
This dissertation is focused on the role RND and MFS efflux pumps play in A. baumannii’s survival strategies under various stress conditions. The first goal was to functionally characterize two putative EmrAB-like MFS pumps, identified as being overproduced in RND efflux deficient strains of A. baumannii, and to gain insight into their role in bacterial stress responses. The second goal was to better understand the substrate recognition of two RND pumps in A. baumannii, AdeG of AdeFGH and AdeJ of AdeIJK, and how specific binding pocket residues impact substrate specificity and export efficiency. I used biochemical methods such as cloning, mutagenesis, antibiotic susceptibility assays, inhibition studies, and fluorescence uptake assays to accomplish these goals. I coupled these biochemical results with genetic and structural analyses to compose a more refined picture of how multidrug efflux protects A. baumannii against extracellular stressors.
I found that while the two newly characterized MFS family efflux pumps, AmfAB and AmfCD, do not contribute to A. baumannii’s antibiotic non-susceptibility phenotypes, they are critical for A. baumannii’s survival under acidic conditions. These results also suggest that these two MFS transporters contribute to stress survival by modification of A. baumannii’s permeability barrier as a method of keeping stressors out of the cell. I also found several AdeG substrate binding pocket residues that are implicated in the export of nalidixic acid, norfloxacin, and the fluorescent probe NPN. Additionally, I used AdeG to characterize kinetic parameters of the putative efflux pump inhibitor, SLUPP_1377. And lastly, we found that the two AdeJ mutants (R701A and F136A) improved the export efficiency of erythromycin, perhaps by better accommodating the antibiotic into the transporter binding pocket. This work provides new insight into the roles multidrug efflux pumps play in bacterial physiology, potential new drug targets which are vital for survival under acidic stress, and informed EPI development strategies from a more detailed model of the AdeG and AdeJ binding pockets
The Pleasure Placebo: The Roles of Religion and Gender in Sexual Enjoyment
No full textThis project focuses on how religious factors relate to heterosexual, cisgender people’s attitudes about gender dynamics and their experiences of sexual enjoyment and well-being. I conduct three related analyses, and each is structured around the premise that religious identities, attitudes, and participation may shape a person’s sexuality in powerful ways. First, I examine how a person’s religious affiliation and level of religiosity predict their level of agreement with male-centric attitudes about sex using the Public Discourse and Ethics Survey (2021). Second, I examine how Christian women’s attitudes about religious gender hierarchies predict their sexual experiences (including orgasm frequency and perceived closeness and communication with their husbands during sex) using the Sexual Satisfaction and Function Survey (2020). Third, I examine how religious affiliation and attendance frequency predict sexual experiences (including orgasm, pleasure, emotional intimacy, and satisfaction) using the National Survey of Sexual Health and Behavior (2014). Woven throughout and alongside these three analyses, I discuss the importance of sexual well-being, theories about gender inequalities in sex, the power of religious dynamics to shape sexual expectations, and the implications of my findings. Taken together, my results show that religious factors are related to sexual attitudes and experiences in important but nuanced ways. Religious participation and affiliation, and Protestantism in particular, are related to larger gender gaps in orgasm and sexual pleasure, and even Christian women who have internalized and embraced a religious gender hierarchy report lower orgasm rates than their more egalitarian counterparts. However, religious engagement with Christianity predicts subjective sexual assessments that are comparable to, or even higher than, non-affiliated women or those who engage with their religious community less often. This suggests that there is a disconnect somewhere between the objective and subjective sexual experiences of Christian women, a phenomenon I call the “pleasure placebo.