Doctor of Philosophy, The Ohio State University, 2018, Vision Science

Introduction: Different components of convergence are tonic convergence, disparity convergence, accommodative convergence and proximal. However, it is not clear whether these different components ultimately draw on similar innervational control. Better understanding of the neurology of convergence eye movements could lead to improvement in interventions for deficits in convergence eye movements. Therefore, the purpose of this dissertation is to investigate the neural basis for convergence eye movements in binocularly normal young adults. Methods: Two approaches were used. First, clinical measurements were used to determine the correlations among accommodative, disparity and proximal convergence eye movements, as well as between proximal convergence and vergence facility. These correlations were used as an index of the extent of overlaps in their neurological control. Second, functional magnetic resonance imaging (fMRI) was performed on a group of adults with normal binocular function as they converged their eyes in response to stimuli for accommodative, disparity, proximal and voluntary convergence eye movements. Results: In the clinical study, stimulus gradient accommodative convergence was negatively correlated with far-near proximal convergence (Spearman's correlation = -0.6111, p < 0.0001). However, the correlation may be at least in part attributable to the inclusion of gradient AC/A as a component of the calculation of far-near proximal convergence. Disparity convergence did not correlate with measures of proximal convergence. Gradient accommodative convergence was not correlated with the amount of disparity convergence in operation (Spearman's correlation = -0.1512, p=0.3649). Finally, proximal convergence was not correlated with vergence facility (Spearman's correlation = +0.1107, p=0.5082 and -0.0149, p=0.9291 for far-near proximal and +2.50D proximal, respectively). In the fMRI study, cluster-based group analysis showed that disparity vergence activated r (open full item for complete abstract)

Doctor of Philosophy, The Ohio State University, 2023, Women's, Gender and Sexuality Studies

This dissertation is a critical consideration of the phrases ‘representation matters' and ‘reading too much into it' and their consequences for the queer reading and interpretive practices of slash fanfiction in light of an influx of LGBT+ mainstream media storytelling over the last 20 years. I contend that, with the advent of more and more dynamic representation of LGBT+ characters and storylines in mainstream media, the expectations for what ‘representations' of queerness may be said to ‘matter' have likewise shifted, resulting in disciplinary action taken against those readers who still insist upon the presence of extracanonical queer potentials in a given media text. By considering the temporal codes through which LGBT+ characters come to matter in media narratives, I will evaluate the limitations of such storytelling and argue that extrapolative fanfiction is a means by which practitioners “[turn] to the fringe of political and cultural production to offset the tyranny of the homonormative,” which has expanded to include many depictions of LGBT+ experience on television and in film (Munoz 2009). By focusing on characters and stories that are frequently taken up by queer fanfiction communities despite lack of intention or confirmation of canonical representation from the media producers, I want to explore the ways in which queer utopian possibilties defy material representation and resist the finitude of contemporary narrative structures. I argue that the queer possibilities of mainstream narrative media objects—even those that are produced in arguably the most normative and rigid production economies—coalesce around characters and in stories that bear important formal and thematic similarities. I will examine queer pairings from several contemporary media narratives, including Supernatural, Sherlock, and the Marvel Cinematic Universe. By considering some of the most popular and prolific queer fanfiction pairings of the last few decades, I explicate these simil (open full item for complete abstract)

Master of Science, The Ohio State University, 2020, Vision Science

Introduction: Convergence insufficiency is a common binocular vision condition that can cause symptoms when reading or doing close work. While there is literature investigating neural activation during disparity vergence in subjects with convergence insufficiency, previous literature has not examined neural activation in convergence insufficiency isolating the components of convergence: accommodative, proximal and disparity vergence. The purpose of this current study is to investigate the blood oxygenation level dependent (BOLD) signal during a functional magnetic resonance imaging (fMRI) scan during accommodative, proximal, and disparity vergence tasks. Methods: Adult subjects (n=6, mean age 23.3 years) with convergence insufficiency (exophoria at least 4 prism diopters greater at near than at distance, receded near point of convergence of 6cm or greater, and insufficient positive fusional vergence). Eligible subjects underwent fMRI scanning that included paradigms stimulating accommodative, proximal, and disparity vergence. Results: Accommodative convergence stimulated regions in the frontal (superior frontal, middle frontal, and precentral gyri), parietal (precuneus, postcentral gyrus), and temporal (middle temporal and inferior temporal gyri) lobes. Proximal convergence stimulated regions in the frontal (superior frontal gyrus), parietal (precuneus, angular gyrus), and temporal (superior temporal gyrus) lobes. Disparity convergence stimulated regions in the occipital lobe. Conclusion: In this subject cohort, there were overlapping areas of activation (in the frontal, parietal, and temporal lobes) observed during accommodative and proximal convergence tasks, while the disparity vergence activation was confined to the occipital lobe. Areas that have been previously linked to convergence, such as the dorsolateral prefrontal cortex, posterior parietal cortex, parietal eye field, frontal and supplementary eye fields, and occipital lobe were observed to be activ (open full item for complete abstract)

Master of Science, The Ohio State University, 2019, Vision Science

Purpose: Convergence Insufficiency (CI) is a common binocular vision disorder that frequently results in symptoms with near work. The purpose of this study was to evaluate the proportion of subjects who could voluntarily converge with and without a target. A second purpose of this study was to determine the sensitivity and specificity of the inability to voluntarily converge for detection of convergence insufficiency and other binocular vision disorders. An additional objective was to compare mean values for NPC, PFV in adults to the limited normative data in the current literature. Methods: Subjects ages 8 years and older with 20/32 or better visual acuity were recruited from visitors at the Center of Science and Industry in Columbus, Ohio. Testing involved administration of the Convergence Insufficiency Symptom Survey (CISS) and evaluation of phoria (Modified Thorington), near point of convergence, positive and negative fusional vergences, accommodative amplitude, visual acuity, and dry autorefraction. Results: Sixty-five participants were enrolled (mean age = 22.3 ± 15.5). Among all participants assessed in this study, 85% were able to voluntarily converge with a target held at 6 cm, while only 55% were able to voluntarily converge their eyes without a target. The majority of participants without signs of CI, AI or significant esophoria were able to converge either voluntarily (75%) or with a target at 6cm (90%). On the other hand, the majority of participants with 2-3 signs of CI were unable to converge voluntarily (38%), but 71% could converge to a target at 6cm. The inability to converge to a target at 6cm was associated with identifying a patient with a 3 sign CI with sensitivity of 0.67 and specificity of 0.89. Conclusion: Inability to converge either voluntarily or to a target at 6cm was associated with signs of CI. Inability to converge to a target identified 2/3 of those with 3 signs of CI.

Master of Science, The Ohio State University, 1924, Graduate School

Master of Arts, The Ohio State University, 1948, Graduate School

Master of Arts, The Ohio State University, 1932, Graduate School

Master of Arts, The Ohio State University, 1960, Graduate School

Master of Arts, The Ohio State University, 1948, Graduate School

MS, University of Cincinnati, 2024, Engineering and Applied Science: Aerospace Engineering

This study underlines the impact of the flow structures on the internal flow field through shape-transitioning ducts with global favorable pressure gradients and local adverse pressure gradients, local to the shape-transitioning geometries. Results are evaluated for convergence apropos of acquisition frequency. This thesis presents preliminary results of flow structure measurement by introducing the structures with a cylindrical bluff body in the cross flow. Structure transport through the two duct configurations studied includes the free jet of a convergent nozzle and through a shape-transitioning nozzle. Particle Image Velocimetry (PIV) was employed in data acquisition considering its substantial spatiotemporal resolution necessary. Findings show that the free jet results are characterized by high-velocity jets, detached velocity deficit region at the tailing edge of the cylinder, and strong velocity gradients due to the shear layers formed between the wake, the jets, and the ambient. On the contrary, flow through the shape transitioning or favorable pressure gradient (FPG) nozzle reflects a well-behaved flow with a low-velocity region attached to the cylinder in most cases. The outcome difference primarily stems from the velocity experienced at the cylinder in each case. An examination of convergence, considering the acquiring frequency of the flow field data, unveiled a weighty impact of acquisition frequency on the results of turbulent flow fields. The ensemble average of the results based on the mathematical computation using analytical methods in the time domain revealed an overall comparable trend in results with notable distinctions in the near wake region. Convergence dependence of results on flow essence emerged with a comparison of the running averages at a point within and outside the wake. In conclusion, it was established that a smaller subset of image pairs drawn from a universal set is ample for effectively capturing the physics of the flow (open full item for complete abstract)

MS, University of Cincinnati, 0, Engineering and Applied Science: Aerospace Engineering

This study underlines the impact of the flow structures on the internal flow field through shape-transitioning ducts with global favorable pressure gradients and local adverse pressure gradients, local to the shape-transitioning geometries. Results are evaluated for convergence apropos of acquisition frequency. This thesis presents preliminary results of flow structure measurement by introducing the structures with a cylindrical bluff body in the cross flow. Structure transport through the two duct configurations studied includes the free jet of a convergent nozzle and through a shape-transitioning nozzle. Particle Image Velocimetry (PIV) was employed in data acquisition considering its substantial spatiotemporal resolution necessary. Findings show that the free jet results are characterized by high-velocity jets, detached velocity deficit region at the tailing edge of the cylinder, and strong velocity gradients due to the shear layers formed between the wake, the jets, and the ambient. On the contrary, flow through the shape transitioning or favorable pressure gradient (FPG) nozzle reflects a well-behaved flow with a low-velocity region attached to the cylinder in most cases. The outcome difference primarily stems from the velocity experienced at the cylinder in each case. An examination of convergence, considering the acquiring frequency of the flow field data, unveiled a weighty impact of acquisition frequency on the results of turbulent flow fields. The ensemble average of the results based on the mathematical computation using analytical methods in the time domain revealed an overall comparable trend in results with notable distinctions in the near wake region. Convergence dependence of results on flow essence emerged with a comparison of the running averages at a point within and outside the wake. In conclusion, it was established that a smaller subset of image pairs drawn from a universal set is ample for effectively capturing the physics of the flow fiel (open full item for complete abstract)

Master of Arts (MA), Bowling Green State University, 2024, Mathematics/Mathematical Statistics

Interface problems frequently appear in numerous physical, biological, and scientific contexts. These problems usually involve differential equations where the input data have discontinuities at one or more interface positions within the solution domain. In this thesis, a numerical method for solving one-dimensional elliptic and parabolic problems with linear and non-linear interface jump conditions at a single interface position is presented. To effectively solve these interface problems, we integrate jump conditions into the numerical method, ensuring that these conditions are met at the interface position. Our approach combines finite difference schemes with a technique termed the $a$-method, specifically devised to address the complexities associated with non-linear interface jump conditions. The convergence behavior of these methods in numerically solving elliptic and parabolic problems with linear interface jump conditions is also examined. Through our numerical examples, we demonstrate that our proposed method achieves an approximate first-order convergence. This occurs because the non-interface grid points exhibit second-order accuracy, while the interface points achieve only first-order accuracy, thereby lowering the overall convergence order to first order when evaluated using the maximum norm criterion.

Doctor of Philosophy, The Ohio State University, 2024, Electrical and Computer Engineering

Online learning encompasses various practical applications where data and learning occur sequentially, such as online optimization, bandit learning, and reinforcement learning. The core challenges of these problems are centered around the efficient utilization of accumulated data. In this thesis, we delve into several intriguing online learning problems with different limitations on the feedback, e.g., the number of oracle feedback is limited per iteration, the feedback is perturbed by an adversary, the reward signal is not available, and on-policy feedback is not available. For each problem, we design efficient algorithms and theoretically capture their performance. Particularly, this thesis makes the following contributions. Firstly, we explore the online nonconvex optimization problem with different oracle feedbacks. To begin with, we consider the settings where only a limited number of oracles are available. We provide the lower bounds of the window-smoothed local regret for the class of linear-span algorithms and show the exact algorithms that achieve the (near-)optimal regret. Subsequently, we study a more challenging setting, where access to only a single oracle is allowed per time step, and take the local regret of the original (i.e., unsmoothed) objective functions as the performance metric. Specifically, we derive lower bounds on the local regret for the class of linear-span algorithms when the single gradient oracle is available and show the (near-)optimal algorithms. Secondly, we investigate the bandit learning under unbounded probabilistic adversarial attack and propose two robust algorithms, median-based $\epsilon$-greedy (med-$\epsilon$-greedy) and median-based exploration-aided UCB (med-E-UCB), and prove that both of them can achieve the logarithmic pseudo-regret, which attains the regret lower bound for the setting without adversarial attack. Our theoretic findings are further validated through computational simulations and experimental result (open full item for complete abstract)

Master of Science, The Ohio State University, 2023, Aero/Astro Engineering

Digital wargames have been used for over half-a-century to better understand the combat environment. From the testing of new strategies and tactics, to testing new technologies in low cost environments before acquisition, digital, and more importantly analytical wargames, are a key asset in any armed forces arsenal. There is a gap in this capability though. Current forms of wargames focus on quantifying the downstream effects of new ideas and technology. This leaves the explanation for how this happened to be described through intuition. Having the ability to statistically prove these intuitions would be invaluable. This is a difficult task to perform given the highly complex, chaotic, and layered system-of-systems that are required to accurately reflect the physical world in the digital world. Thus, a methodology is proposed to provide a means to down select the necessary features to be recorded and modelled. The Advanced Framework for Simulation, Integration, and Modelling (AFSIM) software tool was used to implement and simulate the scenario at hand. The kill chain framework is utilized to craft the recorded features to ensure the necessary information is encoded in the meta-model. Upon collection, features are aggregated into a meta-Markov model representing the probability of transitions between different sets of features over each run. Subsequent analysis can then be performed using centrality measures to identify points in the simulations that could be of interest to analyst.

Master of Science, The Ohio State University, 2023, Vision Science

Purpose: The purpose of this study is to examine all three systems of the near triad and determine if there is a relationship between pupillometry metrics and the presence of binocular vision or accommodation dysfunction after TBI. A secondary objective of this study is to assess the usefulness of a smartphone-based pupillometer as a screening tool for binocular vision and accommodation dysfunction. Methods: Participants were 7 years or older with a lifetime history of at least one TBI. A sensorimotor examination was performed on each participant, including accommodative testing if the participant was under the age of 30 years old. Pupil function was assessed with a clinical pupillometer (RAPDx) and a smartphone pupilometer (Reflex PLR). The average pupil metrics of those with versus without binocular vision dysfunction (based on the sensorimotor exam) were compared using one-way ANCOVA controlling for age. The same analysis was performed for accommodative dysfunction for the group under 30 years old. Receiver operating curves and area under the curve was calculated for every pupil metric for each pupillometer to assess the diagnostic ability of each metric. Results: A total of 28 participants were recruited. Among those under 30 (n = 17), the pupils of those with a binocular vision dysfunction were larger at baseline, took longer to begin constricting, and changed diameter at a higher velocity compared to those without a vision dysfunction. In the over 30 group, the opposite was true. Those with a vision dysfunction had smaller baseline pupil diameters and changed diameter at a slower velocity compared to those without a vision dysfunction. Several pupil metrics measured with the clinical pupillometer demonstrated some ability to identify participants with a vision dysfunction. Only one metric found with the smartphone app showed any ability to identify participants with vision dysfunction. Conclusions: There is a possible association between pupillometry metric (open full item for complete abstract)

Doctor of Philosophy, The Ohio State University, 2023, Electrical and Computer Engineering

Recent decades saw a huge increase in the number of personal devices, wearables, edge devices, etc which led to increased data collection and increased connectivity at the edge. This collected data can be used to make insights about health, the economy, and business and help us make better decisions at the individual, organizational and global levels. With the proliferation of these devices, there are also numerous challenges associated with making use of these devices and the data to train useful models. The challenges could be due to privacy regulations or other constraints determined by the particular learning setup. These constraints make it difficult to extract the required insights from the data and the edge systems. The goal of this thesis is to understand these challenges or resource constraints and develop efficient algorithms that enable us to train models while adhering to the constraints. This thesis makes the following contributions: 1. Propose an efficient algorithm FedCMA for model heterogeneous Federated Learning under resource constraints, showed the convergence and generalization properties, and demonstrated the efficacy against state-of-the-art algorithms in the model heterogeneity setting. 2. Proposed a two-timescale aggregation algorithm that does not require the knowledge of the number of adversaries for defending against Byzantine adversaries in the distributed setup, proved the convergence of the algorithm, and demonstrated the defense against state-of-the-art attacks. 3. We highlight the challenges posed by resource constraints in the Offline Reinforcement Learning setup where the observation space during inference is different from the observation space during training. We propose a simple algorithm STPI (Simultaneous Transfer Policy Iteration) to train the agent to adapt to the changes in the observation space and demonstrated the effectiveness of the algorithm on MuJoCo environments against simple baselines.

Doctor of Philosophy, University of Akron, 2022, Psychology-Industrial/Organizational

This dissertation studies the emergence of Transactive Memory Systems (TMS), specifically the shared knowledge structure component of TMS. To this end, the first part of the project provides in-depth discussions around the theoretical and methodological integration of the TMS and broader teams literatures, the alignment of theoretical and empirical TMS definitions, and the need to study communication in depth as the mechanism through which TMS develops. In the second part of the dissertation, the principles of these discussions were applied to build a computational model of TMS shared knowledge structure emergence in R. Each simulation ran for 100 iterations to study whether communication between agents regarding their areas of expertise resulted in the emergence of the TMS shared knowledge structure. Decision-making and deep learning theories were drawn on to predict that when agents did not have overlap in areas of expertise (had some overlap in areas of expertise), selecting to communicate with the team member thought most likely to be an expert in an information area led to more favorable (less favorable) outcomes than selecting which team member to communicate with randomly. The simulation was repeated for seven learning rates representing how readily agents changed their perceptions about their team members. Results indicated that query and response regarding expertise areas of agents consistently led to the emergence of a TMS shared knowledge structure across conditions. In most instances the pattern of emergence was marked by an initial period of rapid emergence followed by a decrease in the emergence rate. The results supported the hypotheses that when there is no expertise overlap, selecting maximally would lead to a more emerged shared knowledge structure than searching randomly, while the opposite would be the found when there was some expertise overlap. The latter finding challenges the assumption present in the TMS literature that searching for inf (open full item for complete abstract)

Doctor of Philosophy, The Ohio State University, 2022, Industrial and Systems Engineering

Deep Learning Neural Networks (DLNNs) are flexible modeling methods, capable of generating prediction of both continuous and discrete outputs. These methods continue to make large contributions to people's lives. Machine Learning (ML) algorithms are efficient in handling everyday problems, especially big-data ones. DLNNs have a variety of applications, such as recovering disrupted audio files, self-driving cars, YouTube tumblers, and the list goes on. Nonetheless, the performance of DLNNs and ML algorithms, in general, depends upon a collection of choices made by their users. These decisions can be described using factors called “hyperparameters” or “generalized hyperparameters” and further categorized into three groups. We say “generalized” because some of the groups might not conventionally be optimized over. One group defines the structure of a DLNN, for instance, the number of layers, activation functions, and the layer type. The second group relates to the parameters governing the optimization algorithms to derive the weights which minimize the loss function. Some might argue that optimizing over these hyperparameters endangers convergence on training sets for the weight optimization. Yet, here we consider these hyperparameters to be fully adjustable because we argue that fostering test set (unseen data) prediction accuracy is more important than the surrogate goal of achieving convergence on training sets. The third group of hyperparameters relates to controlling data preparation including feature generation and the sampling of training sets. The problem of optimally designing these generalized hyperparameter settings has received relatively little attention. In addition, DLNNs have large numbers of hyperparameters due to their structure. Here, we focus on optimization examples involving eight generalized hyperparameters. The large number of options makes the associate decision problem for DLNN design difficult. The common approaches for this problem include (open full item for complete abstract)

PHD, Kent State University, 2021, College of Arts and Sciences / Department of Political Science

Convergence hypothesis argues that there is a growing similarity among industrialized, labor-importing countries in terms of immigration control policy and integration policy (Hollifield, Martin and Orrenius 2014). This hypothesis also applies in the context of refugee policies that are enacted among these countries. Specifically, two of the most prominent industrialized countries that are built as nations of immigrants and that host refugees are the United States and Canada. Following the logic of convergence hypothesis, the United States and Canada should be similar with regard to immigration control and integration policies. However, Canada is considered an outlier with regard to convergence hypothesis. Convergence hypothesis was descriptive of immigration control in the form of racial exclusion in both countries until 1971 with Canada's adoption of a federal multiculturalism policy. This study tests the convergence hypothesis in the areas of immigration control and integration policies from the United States and Canada by examining policy adoption documents using qualitative content analysis from 1971-2019. The data shows there is a difference in both the inclusivity or exclusivity of the language used by the United States and Canada with regard to refugee policies as well as the frequency of their occurrence. The Canadian turn to multiculturalism was positively associated with more inclusive legislation when examining immigration and refugee policy documents while the United States was more inclusionary with regard to refugee policy. The study also finds there is a clear and definite association between adoption of multiculturalism by Canada and its divergence to being more inclusionary than the United States with regard to refugee resettlement and integration policy. These findings confirm that the adoption of multiculturalism as being associated with Canada's outlier status with regard to convergence hypothesis and demonstrates that industrialized nations can (open full item for complete abstract)

Doctor of Philosophy, Case Western Reserve University, 2022, Mathematics

In this dissertation we implement a method for deriving microscopic rates of con- vergence (ROCs) for determinantal point processes (DPPs) derived from the eigen- values of classic models from random matrix theory (RMT): the Gaussian Unitary Ensemble (GUE), the Laguerre Unitary Ensemble (LUE), and the Jacobi Unitary Ensemble (JUE). Specifically, we obtain bounds on the L^1 -Wasserstein distance be- tween individual point counts for these eigenvalues and the points of their respective limiting DPPs. We obtain two classes of results: one for the GUE bulk eigenvalue DPP and another for the soft edge DPPs for the GUE, LUE, and JUE. In the case of the GUE bulk we obtain ROCs that are O(n^−1 ), a rate that we conjecture to be op- timal. Fascinatingly, our results in this case are powerful enough to yield microscopic information even at some mesoscopic scales. For the three soft edge cases we prove ROCs that are O(n^−2/3 ), which we also conjecture to be optimal. Our soft edge results immediately imply the best-known ROCs for the convergence of the distribution of the largest eigenvalue to the Tracy-Widom distribution for the GUE, LUE, and JUE. Our method hinges on making quantitative estimates of specific trace class norms. The estimates in this work proceed from factoring integral operators and estimating resulting Hilbert–Schmidt norms. The method is adaptable and can in principle be implemented for any DPP to obtain microscopic ROCs to limiting DPPs. However, its practical utility is determined by the ability to effectively estimate the trace class norm that results. We sketch how our method could be implemented in two other cases of interest and highlight where the difficulty lies in completing the method in each case.