MD-SOAR Maryland Shared Open Access Repository
Not a member yet
34521 research outputs found
Sort by
FedMT: Multi-Task Federated Learning with Competitive GPU Resource Sharing
No full textFederated learning (FL) nowadays involves heterogeneous compound learning tasks as cognitive applications’ complexity increases. For example, a self-driving system hosts multiple tasks simultaneously (e.g., detection, classification, segmentation, etc.) and expects FL to retain life-long intelligence involvement. However, our analysis demonstrates that, when deploying compound FL models for multiple training tasks on a GPU, certain issues arise: As different tasks’ skewed data distributions and corresponding models cause highly imbalanced learning workloads, current GPU scheduling methods lack effective resource allocations; Therefore, existing FL schemes, only focusing on heterogeneous data distribution but runtime computing, cannot practically achieve optimally synchronized federation. To address these issues, we propose a full-stack FL optimization scheme to tackle both intra-device GPU scheduling and inter-device FL coordination for multi-task training. Specifically, our works illustrate two key insights in this research domain: Competitive resource sharing is beneficial for parallel model executions, and the proposed concept of “virtual resource” could effectively characterize and guide the practical per-task resource utilization and allocation; Additionally, architectural-level coordination improves FL performance by aligning task workloads with GPU utilization. Our experiments demonstrate that the FL performance could be significantly escalated. Specifically, we observed a 2.16×–2.38× increase in intra-device GPU training throughput and a 2.53×–2.80× boost in inter-device FL coordination efficiency across diverse multi-task scenarios.https://ieeexplore.ieee.org/abstract/document/1112997
Test Validation Pitfalls
No full textThe principles of test validation are well established, but there has been many a recorded slip between accepted theory and acceptable validation studies, or simply the argument that the selection method is appropriate for the situation. This disconnect has legal consequences when the test is implicated in violations of equal employment opportunity laws and regulations. This chapter discusses more than 50 pitfalls in attempting to reach successful validation, illustrated by situations arising in federal court cases. The sections include general problems, as well as those likely to arise during job analysis, content validation strategy, criterion validation strategy, scoring strategy, use of background information, and generalization of validity across selection situations.https://link.springer.com/chapter/10.1007/978-3-319-11143-8_
EVALUATING THE EFFECTS OF THE AFFORDABLE CARE ACT ON HEALTH CARE COVERAGE, UTILIZATION, AND OUTCOMES: A COMPARISON OF FOREIGN-BORN AND US-BORN POPULATIONS
No full textThis dissertation examines the effects of the Affordable Care Act (ACA) Medicaid expansion on insurance coverage, healthcare utilization, and health outcomes among lowincome adults in the United States, with a particular focus on disparities between U.S.- born and foreign-born populations. Using nationally representative survey data and quasiexperimental methods, the three studies presented here trace the pathway from coverage gains to health system engagement and outcomes. The first paper uses American Community Survey data (2011–2019) and a differencein-differences framework with staggered adoption to estimate the impact of Medicaid expansion on insurance coverage. Results show substantial increases in Medicaid enrollment and reductions in uninsurance for both U.S.-born and foreign-born adults, though the latter experienced smaller increases in Medicaid enrollment and larger residual uninsured gaps. Notably, gains were more limited among non-citizen foreign-born adults due to legal status, policy exclusions, administrative barriers, and fear. The second paper evaluates whether coverage improvements translated into increased preventive care by analyzing colorectal and breast cancer screening rates using the Medical Expenditure Panel Survey (MEPS, 2011–2016). Findings indicate that while screening rates improved modestly for U.S.-born adults, foreign-born adults experienced smaller, non-significant gains mirroring the gap in insurance coverage. The third paper investigates the impact of the ACA Medicaid expansion on health outcomes and healthcare expenditures among low-income adults, with a focus on differences by nativity and chronic condition status using MEPS data from 2011 to 2019. This study applies a difference-in-differences framework with interaction terms to estimate changes in health-related quality of life (HRQoL), healthcare utilization (emergency department and hospitalization), and sector-specific spending. The analysis finds that while the expansion significantly reduced out-of-pocket costs and increased Medicaid-financed care—especially emergency services—it did not yield consistent improvements in healthrelated quality of life or reduce hospitalizations overall. An initial decline in mental health scores among foreign-born adults was not robust to sensitivity checks and is interpreted cautiously. Among adults with chronic conditions, avoidable emergency visits declined similarly in both groups, suggesting improvements in care efficiency and financial protection. Together, these studies underscore the complex interaction between immigration status, healthcare policy, and health equity. By applying advanced econometric methods and an intersectional lens, this dissertation highlights both the potential and the limitations of Medicaid expansion in reducing health disparities for immigrant communities. Policy recommendations include improving data collection on immigration status, addressing structural barriers to care, and ensuring inclusive access to public insurance programs
Water Puzzle and Marginal Utility Optimization
No full textWe present a variation of the water puzzle, which is related to a simple model of marginal utility. The problem has an exciting solution and can be extended in several directions.http://arxiv.org/abs/2404.1001
Modeling time scale of integration in equilibrium passive sampling
No full textPassive samplers (PSs) deployed in the field for several months provide a time-averaged measurement of the freely dissolved concentration of pollutants, which is important for assessing ecological exposure and estimating pollutant loads. A comprehensive theoretical modeling assessment of the sampling time scale of integration (TSI) of an equilibrium PS is required to correctly interpret the results. We address this knowledge gap by modeling exchange kinetics of polychlorinated biphenyl congeners in low-density polyethylene (PE) PS based on diffusive transport and first-order kinetics. We evaluate the sampling TSI by analyzing the response of the PS to simulated pulsed concentration increases in the water column that lasted for 1 day in a total sampling period of 90 days. More hydrophobic compounds experience slower transfer into the sampler and show a longer TSI compared with less hydrophobic compounds. Similarly, a thick sampler shows longer TSI than a thinner sampler. The sampling TSI for a typical 25.4 ?m PE sheet ranged widely from 14� days for a dichlorobiphenyl to 43� days for a hexachlorobiphenyl. We show that strategic deployment of a thick and thin passive sampler can be used to narrow the range of TSIs for all congeners and used to simultaneously capture episodic events along with long-term averages.The authors acknowledge financial support from the Department of Defense Strategic Environmental Research and Development Program (Project No. ER-2540; contract # W912HQ14-P-0111).https://academic.oup.com/etc/advance-article/doi/10.1093/etojnl/vgae003/794274
Mononuclear Aluminum–Fluoride Ions, AlFx(⁺/⁻)—Study of Plausible Frameworks of Complexes with Biomolecules and Their In Vitro Toxicity
No full textThe importance of fluorine and aluminum in all aspects of daily life has led to an enormous increase in human exposure to these elements in their various forms. It is therefore important to understand the routes of exposure and to investigate and understand the potential toxicity. Of particular concern are aluminum–fluoride complexes (AlFx), which are able to mimic the natural isostructural phosphate group and influence the activity of numerous essential phosphoryl transferases. Our review of salts of ionic AlFx species, which plausibly form the framework of complexes with biomolecules, revealed that the octahedral configuration of aluminum in the active site of the enzyme is preferred over the trigonal-bipyramidal structure. The effects of varying concentrations of fluoride, aluminum and AlFx—from micromolar to millimolar levels—on the viability and apoptosis rate of THP-1 monocytes were investigated using phosphate buffer solution as a culture media to simulate physiological conditions. Our results suggest that aluminum can reduce the direct toxicity of fluoride through the formation of AlFx. In view of the results found, further in vitro studies are required to clarify the toxicity mechanisms of these species.This work was funded by the Slovenian Research Agency ARIS Grant P1-0045 (Inorganic Chemistry and Technology) and partially by ARIS Grant P3-0086 (Host-Parasite Relationship).https://www.mdpi.com/1420-3049/30/2/38
Predicting the Frequency of Low Cloud Mesoscale Morphologies in Southern Ocean Extratropical Cyclones Using Cloud Controlling Factors
No full textShortwave radiation biases over the Southern Ocean (SO) stem largely from a poor understanding of low clouds in the cold sectors of extratropical cyclones, where rapid transitions between low cloud mesoscale morphologies are frequent. Stratus dominates the poleward regime of the cyclones. It transitions into closed mesoscale cellular convection (MCC) downstream and then to open MCC in the cold sector of cyclones. Clustered and suppressed cumulus are often found in the warm sector. Principal component (PC) analysis is applied to a set of cloud controlling factors to characterize properties of the entire extratropical cyclone that are critical to low cloud mesoscale morphologies. The first two PCs are strongly related to cyclone intensity and sea surface temperature averaged over the cyclone domain, respectively. Daily average insolation at the top of the atmosphere, which has large seasonal and latitudinal variability over the SO, is used as an additional independent predictor. Closed and open MCC are negatively correlated with insolation, while disorganized MCC and clustered cumulus are positively correlated with insolation. In stronger cyclones, closed MCC, open MCC, and clustered cumulus tend to be more frequent, whereas stratus and suppressed cumulus tend to be less common. In cyclones over a colder sea surface, closed MCC and stratus are more abundant, and clustered cumulus and suppressed cumulus are less abundant. These results deepen the current understanding of low cloud processes and provide insights of transitions between morphologies, and thus changes in cloud radiative effects, over the SO in a changing climate.Tong acknowledges funding from the NASA FINESST fellowship (grant: 80NSSC24K0001). The authors want to thank Lynn A. McMurdie for suggestions and discussions throughout the process. There is no financial conflicts of interests for any author. The data used in this study were acquired as part of the mission of NASA’s Earth Science Division and archived and distributed by the Goddard Earth Sciences (GES) Data and Information Services Center (DISC).https://www.authorea.com/doi/full/10.22541/essoar.173724485.59142833?commit=012a6a70e985b5b1c5068ebaeb73a782b46ad2e
Adaptive Control of Dual-Rotor Rotational System with Unknown Geometry and Unknown Inertia
No full textThis paper develops an input-output feedback linearization-based adaptive controller to stabilize and regulate a dual-rotor rotational system (DRRS), whose inertial properties as well as the geometric configuration of rotors are unknown. First, the equations of motion governing the dynamics of DRRS are derived using the Newton-Euler approach. Next, an input-output feedback linearization technique is used to linearize the dynamics from the rotor speeds to the angular position of the system. A finite-time convergent estimator, based on the portion of the DRRS dynamics, is used to update the required parameters in the controller. Finally, the proposed controller is validated in both step and harmonic command-following problems, and the robustness of the controller to the system's parameters is demonstrated.http://arxiv.org/abs/2504.0558
Identifying Flaky Tests in Quantum Code: A Machine Learning Approach
No full textTesting and debugging quantum software pose significant challenges due to the inherent complexities of quantum mechanics, such as superposition and entanglement. One challenge is indeterminacy, a fundamental characteristic of quantum systems, which increases the likelihood of flaky tests in quantum programs. To the best of our knowledge, there is a lack of comprehensive studies on quantum flakiness in the existing literature. In this paper, we present a novel machine learning platform that leverages multiple machine learning models to automatically detect flaky tests in quantum programs. Our evaluation shows that the extreme gradient boosting and decision tree-based models outperform other models (i.e., random forest, k-nearest neighbors, and support vector machine), achieving the highest F1 score and Matthews Correlation Coefficient in a balanced dataset and an imbalanced dataset, respectively. Furthermore, we expand the currently limited dataset for researchers interested in quantum flaky tests. In the future, we plan to explore the development of unsupervised learning techniques to detect and classify quantum flaky tests more effectively. These advancements aim to improve the reliability and robustness of quantum software testing.http://arxiv.org/abs/2502.0447
Capturing Stalinism: Photography, Reality, Manipulation, and Control in Stalin's USSR
No full textDuring Stalin’s rule, the Soviet state, in pursuit of its own survival, sought total control over the lives of its citizens. Many of the methods of pursuing this control have been well explored, particularly the violent repression of any real or perceived dissent. More insidious, however, was the state’s attempts at control through the rewriting of reality. Under Stalinism, thousands of photographs were manipulated, defaced, or partially destroyed, both by the state, and by civilians themselves. Doing so was an act of survival for both parties, but it also contributed to a reality crisis within the USSR. Photographic collections in archives today contain a number of images of denounced individuals who have been painstakingly, fearfully removed from portraits. On their own, these images reveal a society that was in constant paranoia about the presence of certain individuals in photographs. When examined as a pattern, these manipulated photographs suggest that the existence of Stalin’s regime hinged on a contorted relationship to reality itself. This thesis argues that the practice of manipulating and defacing photos was part of a cycle of performance between the state and the people. Through this cycle, multiple, contradictory, simultaneous realities were created, and upheld, by both sides. Understanding this contorted reality is key to understanding what existence was like during Stalinism