MD-SOAR Maryland Shared Open Access Repository
Not a member yet
34521 research outputs found
Sort by
PhysiFi: WiFi Sensing for Monitoring Therapeutic Robotic Systems
No full textThe 23rd International Conference on Pervasive Computing and Communications (PerCom 2025) Washington DC, March 17-21, 2025.Patients recovering from limb-impairing strokes require consistent and precise physical therapy (PT) to regain mobility and functionality. Autonomous rehabilitation robots are increasingly adopted during recovery, offering a scalable solution to reduce the burden on physical therapists while assisting patients in performing prescribed exercises accurately. However, the effectiveness of these treatments often relies on professional supervision to ensure patients follow the robot’s movements properly, which could be challenging considering the ongoing shortage of physical therapists. Current PT monitoring systems primarily rely on camera-based technologies, which usually raise concerns due to potential privacy violations and high deployment costs, or wearable devices that are intrusive and uncomfortable for patients. To address these limitations, we propose PhysiFi, a novel approach that leverages ubiquitous WiFi signals available in most indoor environments, such as homes, rehabilitation centers, and assisted living facilities. By analyzing Channel State Information (CSI) from ambient WiFi signals and employing deep learning models, PhysiFi can track and recognize exercises performed by patients with rehabilitation robots. Our experiments demonstrate that PhysiFi can accurately identify prescribed exercises and evaluate whether patients are following the robot’s movements correctly, providing a non-intrusive, privacy-preserving, and costeffective alternative for monitoring physical therapy sessions.https://ebulutvcu.github.io/PhysiFi-WiSense25.pd
How ticket-splitting voters could shape the 2026 midterms
No full textDemocrats are desperate to retake control of Congress; Republicans want to keep it. To win, it helps to know what kind of voter is willing to cross party lines.http://theconversation.com/how-ticket-splitting-voters-could-shape-the-2026-midterms-24601
Differential Outcomes and Behavioral Units
No full textOur behavioral units include stimulus classes and response classes. Peter Urcuioli's differential-outcomes research implies they should extend to the third term of the three-term contingency. Classes of consequences come in several varieties (e.g., conditional reinforcers, tokens), but our vocabulary does not coherently organize them. They are differentiated not only by physical properties such as type, location and duration but also by the schedule contingencies in which they participate. We consider units ranging from the physical and chemical sciences to those based on the particular history of life on earth. The latter include biology, sociology, linguistics, and our own behavior analysis. Scientific units are typically nested (e.g., atoms within molecules; cells within organs; organisms within species). Comparing our units with those from other taxonomies raises questions about their emergence and evolution and their shared properties across levels of nesting (e.g., species within genus; subclasses within higher-order operants; phonemes within words). Emergence necessarily occurs when higher-order units have functions not shared with their lower-order constituents. These nested and multi-leveled behavior classes challenge single-level views, such as metaphorical accounts of behavior as a totality contained within a pie with slices corresponding to behavior classes matched to their outcomes.https://onlinelibrary.wiley.com/doi/10.1002/jeab.7001
Looking at infrared background radiation anisotropies with Spitzer: large scale anisotropies and their implications
No full textWe use Spitzer/IRAC deep exposure data covering two significantly larger than before sky areas to construct maps suitable for evaluating source-subtracted fluctuations in the cosmic infrared background (CIB). The maps are constructed using the self-calibration methodology eliminating artifacts to sufficient accuracy and subset maps are selected in each area containing approximately uniform exposures. These maps are clipped and removed of known sources and then Fourier transformed to probe the CIB anisotropies to new larger scales. The power spectrum of the resultant CIB anisotropies is measured from the data to >1 degree revealing the component well above that from remaining known galaxies on scales >1 arcmin. The fluctuations are demonstrated to be free of Galactic and Solar System foreground contributions out to the largest scales measured. We discuss the proposed theories for the origin of the excess CIB anisotropies in light of the new data. Out of these, the model where the CIB fluctuation excess originates from the granulation power due to LIGO-observed primordial black holes as dark matter appears most successful in accounting for all observations related to the measured CIB power amplitude and spatial structure, including the measured coherence between the CIB and unresolved cosmic X-ray background (CXB). Finally we point out the use of the data to probe the CIB-CXB cross-power to new scales and higher accuracy. We also discuss the synergy of these data with future CIB programs at shorter near-IR wavelengths with deep wide surveys and sub-arcsecond angular resolution as provided by Euclid and Roman space missions.Work by A. K. and R. G. A. was supported by NASA under award number 80GSFC24M0006. The authors acknowledge support from NASA award 80NSSC22K0621 “Precision measurement of sourcesubtracted cosmic infrared background from new Spitzer data”.https://iopscience.iop.org/article/10.3847/2041-8213/adad5
Competence Measure Enhanced Ensemble Learning Voting Schemes
No full textEnsemble Learning Methods Ensemble learning methods use the predictions of multiple classifier models. – A well-formed ensemble should be formed from classifiers with various assumptions, e.g., differing underlying training data, feature space selection, and therefore decision boundaries. A voting scheme is used to weigh the decisions of the individual classifier models to determine how they may be combined, fused, or selected among to predict class. – Voting schemes often consider individual reported classifier confidence in predictions. Complementary features, class representation, and training data distribution across the classifiers are to an advantage, but are not being fully exploited with existing schema. Network approaches attempting to learn the complementary traits of classifiers may result in loss of explainability to end users.DATAWorks 2025 Alexandria VA Contributed Session 5C Advancing T&E of Emerging and Prevalent Technologies Improving Quality of T&E 24 April 2025https://dataworks.testscience.org/wp-content/uploads/formidable/23/Thrs_5C_McFadden.pd
LLM-ProS: Analyzing Large Language Models' Performance in Competitive Problem Solving
No full textLLM4Code 2025: The Second International Workshop on Large Language Models for CodeThe rapid advancement of large language models has opened new avenues for automating complex problem-solving tasks such as algorithmic coding and competitive programming. This paper introduces a novel evaluation technique, LLM-ProS, to assess the performance of state-of-the-art LLMs on International Collegiate Programming Contest (ICPC) problems. Using a curated dataset of 166 World Finals problems from 2011 to 2024, we benchmark the models' reasoning, accuracy, and efficiency. We evaluate the five models-GPT-4o, Mistral Large, Llama-3.1-405B, and the o1 family, consisting of o1-mini and o1-preview, across critical metrics like correctness, resource utilization, and response calibration. Our results reveal significant differences in the models' abilities to generalize, adapt, and solve novel problems. We also investigated the impact of training methodologies, dataset contamination, and chain-of-thought reasoning on model performance. The findings provide new insights into optimizing LLMs for algorithmic tasks, highlighting both strengths and limitations of current models.http://arxiv.org/abs/2502.0435
ANNS: An Intelligent Advanced Non-Convex Non-Smooth Scheme for IRS-Aided Next Generation Mobile Communication Networks
No full textEnhancing the communication rate and quality has become the primary goal for the development of next-generation mobile communication networks, and traditional techniques such as MIMO and increasing the transmit power of the base station (BS) have not achieved a leapfrog effect. The emergence of Intelligent Reflective Surfaces (IRS) provides more reliable technical support for providing high-energy and high-rate communications. However, IRS-aided joint optimization with communication rate and quality of service constraints is a non-convex non-smooth optimization problem, and the optimal global solution cannot be obtained due to its computational complexity. In this paper, we propose an intelligent Advanced Non-convex Non-smooth Scheme (ANNS) in IRS-aided Next Generation Mobile Communication Networks for making the transmission rate of mission communication and communication quality effective. To ensure that the inequality constraints in the joint optimization problem are not violated and the equation constraints are satisfied, a hybrid deep reinforcement learning and data-experience-driven constraint security layer network is proposed, which maps the constraint violations into the safe feasible domain by mapping the constraint variables into the constraint variables mapping method, and the convergence of the algorithm is theoretically demonstrated. Experimental results show that the proposed ANNS performs superior optimization compared to SAC, DDPG, and A2C for solving non-convex non-smooth problems. The proposed ANNS has the potential to be generalized to other mobile computing applications with non-convex non-smooth characteristics.This work is partially supported by the National Natural Science Foundation of China (Nos. 62462002), and partially supported by the Natural Science Foundation of Guangxi, China (Nos. 2025GXNSFAA069958, 2025GXNSFBA069394)https://ieeexplore.ieee.org/document/10959110
Trump’s executive order to dismantle the Education Department was inspired by the Heritage Foundation’s decades-long disapproval of the agency
No full textThe conservative think tank’s efforts to dismantle the Education Department go back more than 40 years, a scholar writes.http://theconversation.com/trumps-executive-order-to-dismantle-the-education-department-was-inspired-by-the-heritage-foundations-decades-long-disapproval-of-the-agency-25060
The Hidden Crisis in Baltimore's Public Safety Pension System: The Impact of Declining Active Membership on Baltimore's Public Safety Pension System and the City Budget
No full textPublic Scholarship Project submitted to the College of Public Affairs of The University of
Baltimore in partial fulfillment of
the requirements for the degree
of Doctor of Public Administration.D.P.A.-- The University of
Baltimore, 2025This study examines the critical and often overlooked issue of declining active membership in the Baltimore City Fire and Police Employees’ Retirement System (BCFPERS, F&P) and its far-reaching implications for the system's long-term sustainability and the city budget. Guided by the central research question: How does declining active membership impact the long-term sustainability of BCFPERS and the resulting budgetary implications for city government?—the research employs a mixed-methods approach, combining qualitative and quantitative data analysis.
By analyzing demographic shifts, financial data, and stakeholder perspectives, this study identifies structural vulnerabilities threatening the system’s financial stability. To increase public understanding and accessibility, the project includes an audio component: Silent Alarm: Public Safety Pensions, a podcast series featuring candid discussions with BCFPERS leadership, investment consultants, and actuaries on pension challenges and actionable solutions.
A companion website provides access to podcast episodes along with visual data tools (charts and graphs) that illustrate funding ratios, membership trends, and budgetary impacts. These resources simplify complex financial concepts, summarize findings, and foster transparency. Ultimately, this study provides targeted recommendations to strengthen the long-term sustainability of BCFPERS and contributes to the broader discourse on public sector pensions, fiscal responsibility, and urban fiscal resilience.https://ziginc.com/silent-alarm
Toward Generative 6G Simulation: An Experimental Multi-Agent LLM and ns-3 Integration
No full textThe move toward open Sixth-Generation (6G) networks necessitates a novel approach to full-stack simulation environments for evaluating complex technology developments before prototyping and real-world implementation. This paper introduces an innovative approach\footnote{A lightweight, mock version of the code is available on GitHub at that combines a multi-agent framework with the Network Simulator 3 (ns-3) to automate and optimize the generation, debugging, execution, and analysis of complex 5G network scenarios. Our framework orchestrates a suite of specialized agents -- namely, the Simulation Generation Agent, Test Designer Agent, Test Executor Agent, and Result Interpretation Agent -- using advanced LangChain coordination. The Simulation Generation Agent employs a structured chain-of-thought (CoT) reasoning process, leveraging LLMs and retrieval-augmented generation (RAG) to translate natural language simulation specifications into precise ns-3 scripts. Concurrently, the Test Designer Agent generates comprehensive automated test suites by integrating knowledge retrieval techniques with dynamic test case synthesis. The Test Executor Agent dynamically deploys and runs simulations, managing dependencies and parsing detailed performance metrics. At the same time, the Result Interpretation Agent utilizes LLM-driven analysis to extract actionable insights from the simulation outputs. By integrating external resources such as library documentation and ns-3 testing frameworks, our experimental approach can enhance simulation accuracy and adaptability, reducing reliance on extensive programming expertise. A detailed case study using the ns-3 5G-LENA module validates the effectiveness of the proposed approach. The code generation process converges in an average of 1.8 iterations, has a syntax error rate of 17.0%, a mean response time of 7.3 seconds, and receives a human evaluation score of 7.5.http://arxiv.org/abs/2503.1340