33760 research outputs found
Sort by
Toehold Switch Sensor for miRNA Cancer Biomarker
No full textColorectal cancer (CRC) is the third most common cancer in the United States, yet existing screening and diagnostic methods remain expensive, invasive, and require many resources, discouraging early testing. Abnormal microRNA (miRNA) expression has been shown to be a cancer biomarker, with miR-29a being upregulated in the blood of CRC patients. To address this gap, we sought to develop a low-cost, portable, and minimally invasive assay for CRC miR-29a detection in blood plasma. We first had to construct our toehold switch. The toehold switch was assembled using Gibson Assembly and transformed into E. coli, isolated through plasmid miniprep, and verified through Oxford Nanopore long-read. After sequencing the biosensor, we determined we have successfully assembled the toehold switch and will begin testing of its efficacy. We are now currently testing the sensitivity and specificity of our toehold switch construct reporter for our target miR-29a biomarker. Afterwards, we will optimize the performance of our toehold switch reporter & integrate it into a one pot TRAP system, a Thermally Responsive Alkane Partition System, to streamline the reactions leading to Green Fluorescent Protein (GFP) expression in a positive test. However, due to time constraints we were unable to complete all the cell-free testing of our miR-29a and toehold switch and get to the TRAP system but we have gotten paradoxical results as our different constructs with and without the presence of miRNA had values that were significantly different, potentially indicating the success of our toehold
Pine Street's Preparedness for Extreme Weather and Heat
No full textThis community-level resilience report card examines the preparedness of the Pine Street neighborhood in Cambridge, MD to threats like severe weather and extreme heat. After surveying community members to identify priority concerns in the neighborhood, indicators to assess threats and resilience around those concerns were selected to give the community a preparedness score. This report card recommends individual and community-wide actions that can reduce vulnerability in areas of concern.https://ian.umces.edu/site/assets/files/32738/pine-streets-preparedness-for-extreme-weather-and-heat.pd
Multilevel Racism and Discrimination and Cardiovascular Disease and Related Biopsychosocial Mechanisms: An Integrated Scoping and Literature Review and Future Research Agenda
No full textPURPOSE OF REVIEW: In the last two decades, empirical research has significantly advanced our understanding of the link
between discrimination and cardiovascular disease (CVD). This integrated scoping and narrative literature review delineates
the extant peer-reviewed research on discrimination and clinical and subclinical CVD in samples that include Black adults,
using a multilevel conceptualization of race-related discrimination and racism. We also identify potential intermediary
mechanisms in the racism-CVD relationship and propose a comprehensive future research agenda.
RECENT FINDINGS: Using the Population, Exposure and Outcome framework and PRISMA guidelines, we identified 37 empirical reports for inclusion drawn from 1900 to 2024. The bulk of the literature has focused on discrimination and racism that
occurs at the interpersonal level (28 studies), while a smaller but growing body of work has examined cultural (5 studies) or
institutional and structural-level racism and discrimination (4 studies) in relation to CVD risk. The majority of these studies
show that greater exposure to discrimination or racism is associated with increased clinical or subclinical CVD risk. Potential pathways include societal, environmental, psychological, and biological factors; however, few studies have conducted
formal tests of mediation.
SUMMARY: The literature suggests robust relations of multilevel racism and discrimination to manifestations of CVD across
diverse exposure and outcome measures in Black adults. Our recommendations to eliminate cardiovascular health inequities in Black communities include enhancing academic scholarship training, securing targeted and protected funding, and
adopting more robust methodological approaches.Beatty Moody received funding support from P30 AG028747.https://doi.org/10.1007/s11886-025-02238-
Build A Track: How to Learn and Teach Resource Sharing within the MLIS Curriculum
No full textThere is a need to prepare future resource sharing professionals, but this specialization is not typically part of the MLIS curriculum. To address this gap the authors propose building a track of courses that could prepare future librarians for a career in resource sharing when paired with learning outside the classroom. They demonstrate the feasibility of this approach by mapping RUSA’s Professional Competencies for Resource Sharing Practitioners (2024) and Guidelines for Resource Sharing Operations Management (2022) to the MLIS curriculum at the University of Maryland’s College of Information. Core courses and electives are evaluated to identify which competencies were being met through the curriculum, where hands-on learning or resource sharing community resources would be more effective, and courses that could feature resource sharing to further strengthen the pipeline.https://doi.org/10.1080/26915979.2025.259195
2025 Chesapeake Bay & Watershed Report Card
No full textThis report card provides a transparent, timely, and geographically detailed assessment of Chesapeake Bay and its Watershed. Since 2016, UMCES has engaged stakeholders throughout the watershed to transform the report card into an evaluation of the Chesapeake Watershed health. Watershed health includes traditional ecological indicators, but also economic and societal indicators. This is the sixth year the watershed has been scored, and one new ecological indicator, Temperature Stress, has been added.
Overall, the Chesapeake Watershed scored 57%, a C+. There were five ecological indicators, four economic indicators, and three societal indicators.
Overall, Chesapeake Bay scored 50%, a C, in 2025. This was a slight downturn from the previous year, but still a significantly improving long-term trend.https://ian.umces.edu/site/assets/files/32674/2025-chesapeake-bay-watershed-report-card.pd
Practical Multiparty Protocols From Lattice Assumptions: Threshold Signatures, Oblivious Pseudorandom Functions, And More
No full textLattice-based cryptography has emerged as the most dominant replacement candidate for the next generation of post-quantum cryptographic tools. With their operational simplicity while allowing advanced functionality, these protocols lead the majority of post-quantum standardization efforts and motivate a great chunk of current research to realize advanced trusted communication models. However, lattices' greatest asset is also their greatest curse. The applicability of advanced functionality motivates protocols with multiple computing parties while the assumptions that make lattice protocols secure in the first place hate settings where secrets are distributed.
In this work we try to alleviate this issue by building practical lattice-based multiparty protocols. First we propose the first known concrete lattice-based threshold signature scheme with distributed key generation to demonstrate practicality. Second, we look at a different type of protocol, namely verifiable oblivious pseudorandom functions, and propose a practical version of an existing protocol through different analysis techniques while also giving the first lattice-based threshold versions of such protocols. Using these techniques, we then rebuild our threshold signature scheme and show a concretely efficient threshold signature that simultaneously provides additional desirable properties like identifiability and non-interactivity. Finally, we look at the possibility of asymmetric outsourced computation and formalize the classic notion of augmented password-protected threshold signatures in a more practicality friendly manner and construct the first lattice-based augmented password-protected threshold signature scheme.
All of these works act as building blocks for more complicated protocols and share similar analysis techniques and solutions to problems specific to the distributed setting. This commonality indicates that it is not only the assumptions that we need to revisit but also how we think about security in general as part of preparing cryptography for its post-quantum era
Multi-Agent Autonomous Decision Making in Artificial Intelligence
No full textMulti-Agent Autonomous Decision Making, especially Multi-Agent Reinforcement Learning (MARL), is an emerging area of Artificial Intelligence (AI) where autonomous agents interact with each other, fostering competition and/or cooperation. These AI agents can be useful to solve real world problems like Augmented Reality, Recommender Systems, Supply Chain Orchestration, Climate Conservation, Self-Driving Cars, Sports, Interdiction Games and Real-time Guidance of Cooking, Education, Manufacturing and Robotic Tasks. Challenges of AI Agents include efficiently scaling to multiple agents, solving coordination problems and understanding agentic behavior.
My Ph.D. thesis has the objective to develop and deploy efficient Multi-Agent AI algorithms, for real-world decision making problems. To begin with, a multi-agent approach can be used to model the Human-AI Alignment problem, a major challenge to rapidly deploy AI models. Mis-alignment challenges exist in current AI models like ChatGPT which face serious challenges to plan or reason like performing a 4-digit multiplication of two integers. Concepts from MARL like ad-hoc coordination help humans and autonomous agents to communicate the goals of each planned step correctly and explicitly reason about the strategies humans may utilize when attempting to shape the behavior of AI model agents. These kinds of communications are not often efficient or robust with increasing scale of AI Agents and need theoretical results for efficient Agentic behavior. I have provided formal guarantees for successful and reliable cooperation of AI agents with populations of socially intelligent agents, defining agentic behavior with game-theoretic notions of consistency and compatibility. AI agents cooperate in the above settings with populations of socially intelligent agents that are individually rational while also reliably coordinating with other group members in a general-sum Bayesian game. The AI agents face challenges generalizing from previous interactions that can help them to cooperate with a new partner drawn from such populations. It is theoretically shown that just these assumptions are insufficient to select an AI agent’s strategy that achieves zero-shot coordination with any member in the socially intelligent population, which can be addressed by a proven upper bound on the sample complexity to learn a successful cooperation strategy, based on observing interaction among members of the target population. Lower bounds are derived to show when the multi-agent cooperation setting is needed with respect to the populations’ trajectories, the state space and the length of the learning episodes. These bounds under the assumption of consistency and compatibility are proven to be stronger than a “naive” reduction of this cooperation problem to one of Imitation Learning.
My thesis then shows that such collaborations of AI Agents in Alignment with human goals can have real-world applications like with Augmented Reality and Self-Driving Cars. Multimodal vision-language AI Agents can assist humans proactively by determining when and how the AI Agent will autonomously intervene in real-time to cooperatively solve day-to-day tasks. Augmented Reality (AR) gadgets with distributed edge computing use cases, be it a smartphone or a wearable device, can lead to a major improvement of the user experience in solving procedural day-to-day tasks by introducing egocentric multimodal (audio and video) observational capabilities to AI agents. These AR capabilities help the AI Agents to see and listen to users' actions, thus relating to multimodal capabilities of human users. Current AI Agents, be it Large Language Models (LLMs) or Multimodal Vision-Language Models (VLMs) are mostly reactive in nature, where the AI models cannot take an action without waiting for the human user’s vision-language prompts. Proactivity of AI Agents helps the human users to detect and correct any task mistakes by providing more autonomous assistance, encouraging users when they do tasks correctly or simply engaging in conversation with users - akin to a human teaching or helping another human user. I have created a YET to Intervene (YETI) multimodal agent that focuses on the research question of identifying circumstances in real-time that may require the AI agent to proactively intervene. My trained YETI agent can understand when it can intervene in a conversation with human users to help them correct mistakes on tasks, like cooking, using Augmented Reality. YETI learns scene understanding signals based on interpretable notions of Structural Similarity (SSIM) on consecutive observed video frames. It also learns the alignment signal to identify if the video frames corresponding to users' actions on the task are consistent with their expected actions. These signals are used by the AI Agent to determine when it should proactively intervene. I compare the YETI results on the instances of proactive intervention to the HoloAssist multimodal benchmark for an expert agent guiding a user to complete procedural tasks. Control problems for autonomous AI agents, especially safety-critical applications such as autonomous vehicle control, require robust decision making frameworks to ensure safe navigation in such complex and dynamic environments. This necessitates approaches such as Agentic Model Predictive Control (MPC), which can anticipate future problems and plan for them accordingly. A novel framework has been introduced that integrates MPC with Multimodal VLMs in order to enhance the ability of autonomous vehicles to navigate and respond to real-world scenarios with the ability to take fine-grained actions.
Multi-Agent AI can be pervasive in real world applications, given the foundation of humans and other technological agents to interact with each other, strategize and perform a task. In my thesis, I show that MARL can help to strategize mitigation strategies for climate conservation problems like deforestation mitigation by improving the prediction of deforestation hotspots in Indonesia, one of the two major rain forests in the world. I also share the modeling of another application of multi-agent AI collaboration in Supply Chain Orchestration, creating a simulated environment that is cognizant to seasonal demand and cold chains with improved exploration of strategies to maximize profit. I have created a new intrinsic reward signal, helping to save unnecessary interactions among AI Agents planning inventory in Supply Chain warehouses.
These real-world applications motivate the need to understand why the AI Agents behave the way they do which is addressed with Explainable AI (XAI) agents by addressing the question of which XAI methods should be recommended, subject to user agent goals. Explaining the behavior of AI models becomes important in the context of different factors including their training and inference speed that can determine end-users preferring an AI model over another. MARL has been applied to Explainable AI (XAI) problems using a Multi-Agent RecSys to recommend Explainable AI (XAI) outputs for different AI models that can serve the objectives of the model’s users for building trustworthy safe AI. Goal-Conditioned RL can be applied to model AI users learning XAI outcomes as per their preferences. Research on learning to visualize semantic representations satisfying user objectives provides motivation to improve the visualization of XAI methods satisfying the objectives of different users as targets with MARL representations. To represent MARL targets, much of the control problem can be abstracted for deployment in real-world settings like interdiction games, with a much simpler game theoretic problem. Multi-Agent AI algorithms also help to prune AI model parameters across model layers for efficient learning.
As with humans, a large number of AI agents can take a long time to learn strategies jointly. Multi-Agent RL can be pretty slow with increasing scale of agents. To address this, it is shown that the JAXMARL library leverages JAX-enabled hardware acceleration that can make it 12,500x faster over existing libraries in 8 popular MARL environments. The effectiveness of AI Agents can be improved by a combination of the proposed Multi-Agent Reinforcement Learning, Imitation Learning, Model Predictive Control, and Computational Game Theoretic algorithms in solving problems in real world and simulation environments
PHYSICS OF RELATIVISTICALLY SELF-FOCUSED LASER PROPAGATION IN PLASMA
No full textThe capabilities and achievable intensities of laser systems has grown by leaps and bounds in the past several decades especially since the development of chirped pulse amplification, and the typical behavior of such light has diverged greatly from the well-understood linear optics to new highly nonlinear optical regimes. This has opened the door to a plethora of new phenomena and applications including long-distance filamentation, high harmonic generation, and many varieties of laser-plasma acceleration. Of particular interest to this dissertation is the physics of relativistically self-focused laser pulse propagation through plasma such as during laser-wakefield acceleration. The objective of this dissertation is to optimize relativistically self-focused laser pulses in plasma for kHz laser wakefield acceleration, investigate the emergence and role of spatio-temporal optical vortices in relativistic filamentation, and to further isolate and analyze the interaction between such vortices and plasma. We demonstrate the ability to accelerate electron bunches up to 15 MeV using few-cycle, mJ-scale pulses at a kHz repetition rate by taking advantage of relativistic self-focusing in near-density hydrogen plasma. Laser polarization is shown in both particle-in-cell simulations and in experiment to affect the energy spread, charge, and divergence of accelerated electrons through CEP-driven asymmetric driving of the wakes; we find that circular polarization resulted in the most monoenergetic and collimated electron beams. Furthermore, we investigate the emergence of spatio-temporal optical vortices that arise in relativistic filamentation/self-focused propagation in plasma using particle-in-cell simulations and compare them with non-relativistic filamentation. We find that these vortices are fundamental features of such propagation and mediate the intrapulse flow of energy during filamentation. Lastly, we isolate these spatio-temporal optical vortices (STOVs) and examine their propagation in plasma in the linear regime. As these vortices are known to carry transverse orbital angular momentum, we use particle-in-cell simulations to directly track the evolution of angular momentum in both the fields and particles during the STOVs’ propagation in plasma. Consistent with our group’s previously derived analytic theory, STOVs share angular momentum with the medium in a polariton structure
SPARSE FEATURE SELECTION AND REGIME IDENTIFICATION FOR ADVANCED PROGNOSTIC METHODS FOR COMPLEX SYSTEMS
No full textThis dissertation introduces two novel methods—Sparse Multivariate Functional Fusion Predictors (SMFFP) and Data-Driven Regime Clustering with Spectral Approximation (DRC-SA)—to enhance interpretable prognostic modeling for complex systems, particularly aircraft engines and Switch-Mode Power Supplies (SMPS). While traditional black-box approaches, such as deep learning, often achieve high predictive accuracy, their lack of interpretability limits their utility in safety-critical applications. In contrast, SMFFP and DRC-SA provide a transparent, data-driven framework capable of capturing system degradation patterns with both precision and clarity.
The SMFFP method integrates sparse multivariate functional predictions with Koopman operator theory to create interpretable models. By mapping nonlinear system dynamics into a linear framework, Koopman theory enables the identification of key observables that characterize system behavior. Additionally, sparse feature selection techniques within SMFFP further enhance model clarity by isolating the most critical predictors while maintaining competitive predictive performance, even in data-constrained environments.
Complementing SMFFP, the DRC-SA method identifies operational regimes critical for accurate degradation predictions. Using spectral clustering with the Nyström approximation, DRC-SA effectively clusters spatio-temporal patterns under both normal and anomalous conditions. Rather than treating an operational regime as purely normal or anomalous, DRC-SA captures regime-specific variability, allowing it to classify regimes that may exhibit both normal and anomalous health conditions as either distinct sub-regimes or variations of a single regime, depending on the persistence and structural differences of the anomalies. This facilitates detailed classification of operational regimes, ensuring that predictive insights align with real-world operational changes and enabling the early detection of system degradation.
Together, SMFFP and DRC-SA provide a robust and interpretable framework for Prognostics and Health Management (PHM) in complex systems. These methods address the critical need for predictive maintenance solutions that prioritize transparency and reliability in safety-critical applications
The Ethical and Moral Dilemma of Artificial Intelligence
No full textWhen the ALA updated its core competencies in 2022, it was the first time in the library profession to intentionally incorporate “concepts of social justice, equity, diversity, and inclusion.” Despite this, Galleries, Libraries, Archives, and Museums (GLAMs) have made an effort in recent years to adopt and implement machine learning and artificial intelligence. On one hand, GLAMs are known to be learning institutions and are touchpoints for communities to usher in breakthrough or new technologies. However, implementing A.I. and machine learning into these institutions is directly contradictory to the core competencies and would sever the trust gained with marginalized and underrepresented communities. Stanford tested three popular AI models—ChatGPT, Google AI, and RoBERTa—to examine whether they responded differently to identical crimes when the defendants were of different races, one white and one black. The language models exhibited attitudes “even more negative than the most negative experimentally recorded human attitudes about African Americans.” (p. 149) Overall, the AI was more likely to convict the black defendant, sentence them to prison, and even opt for the death penalty despite having committed the same crime. In the United States, the total adoption rate of artificial intelligence in libraries currently sits at 21%. While that may seem low, awareness and use are only going to increase as GLAMs become ever more susceptible to budget cuts, poor working conditions, and divestment in public initiatives. The findings here show that using AI is only going to further marginalize underrepresented groups while making the workflow marginally easier for those who already benefit from privileged conditions.https://drive.google.com/file/d/1Xz62MB5b-5_sInamC2GIMLpTSqCLViMt/view?usp=drive_lin