George Mason University

NSF Awards · FY 2025 · 2025-09

The aim of this project is to advance disaster decision making, risk assessment, and management science by incorporating access to shelters and the ability to seek shelter into the national disaster risk assessment. Management decisions on resource allocation and emergency preparedness planning are hindered when disaster risk is not well understood. The existing tools and resources available to emergency managers often lack measures of shelter accessibility and the ability to seek shelter, which can escalate natural disasters into human disasters. By advancing existing measures, this translational project equips emergency managers with the knowledge and tools needed to cope proactively with disasters such as wildfires, hurricanes, tornadoes, and coastal floods. The potential societal benefits in this project include engaging emergency directors, planners, and operators to minimize redundancy and tool fatigue and ensure that outcomes align with their needs, and improving the well-being and survival of populations by identifying gaps in shelter access and prioritizing the allocation of shelter and mobility resources. This effort is guided by a vision of improving disaster risk understanding within a framework that integrates community resources and capabilities into risk assessment, management, and decision making. The research team achieves this goal through three research activities, co-produced in close collaboration with emergency managers across the nation. First, the research team develops a comprehensive, risk-based national shelter accessibility model to advance the state of the art in shelter accessibility measurement by accounting for both the availability and accessibility of shelters, as not all shelters remain functional during disasters. Second, the research augments existing national measures by integrating shelter accessibility and evacuation capabilities to enhance both short-term and long-term emergency management decisions. Third, the researchers create a science-informed decision-making tool to test risk perception and decision making in emergency management, enabling emergency directors, planners, and operators to explore how short-term and long-term strategies can provide evacuees with a better chance to survive. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-09

Prosthetic limbs designed for physical activity, such as running blades, allow children to participate in sports and stay active. However, these specialized prostheses often are expensive and not always covered by insurance, making them difficult for many families to afford. Emerging technologies, such as 3D printing and cloud-based design tools, offer the potential to lower costs and to customize prostheses as children grow. Nevertheless, these innovations remain underused. This project will ascertain what children need from their prostheses by examining how their motivation to be active, body size, and movement type influence prosthetic performance. It will compare how advanced 3D-printed prostheses perform relative to traditional models under real-world conditions. Ultimately, this research will make high-performance prosthetic limbs more affordable, accessible, and tailored to the needs of active children. In addition, the project will create research and educational opportunities for students, introducing them to advanced manufacturing techniques, biomechanics, and patient-centered design, which will foster interest in STEM fields and help inspire future biomedical engineers. Recent advances in composite additive manufacturing and cloud computing have created new opportunities for the rapid, cost-effective production of complex, high-performance components. These technologies are well-suited to improve the design and fabrication of physical activity enabling prostheses (PAEPs) for children, offering scalable customization to accommodate growth and varied activity demands. Despite this potential, they remain underutilized in pediatric prosthetic development. This project addresses this gap by integrating patient-centered insights, mechanical testing, and advanced manufacturing to define design criteria for pediatric running-specific prostheses (RSPs). Research activities include the collection multidimensional data through surveys of children with lower-limb absence and their parents and clinicians. Statistical models will identify key predictors of prosthetic satisfaction and physical activity participation, while thematic analysis of open-ended responses will highlight subjective barriers and facilitators. Reported activities will be deconstructed into their underlying biomechanical demands and the mechanical behavior of three commercially available pediatric RSPs under static and dynamic loading will be evaluated. These tests will link stiffness and fatigue performance to user-specific anthropometric and movement data, informing predictive models of prosthetic response. Guided by these models, custom PAEPs using continuous fiber fused filament fabrication (CF-FFF), enhanced through Additive Fusion Technology (AFT) to improve structural integrity will be designed and fabricated. Mechanical evaluation of these devices will include standard tests, 3D digital image correlation for strain mapping, and micro-computed tomography for internal fiber analysis. The project will generate open-access models and fabrication protocols that advance personalized, high-performance pediatric PAEPs. Additionally, it will support STEM outreach through summer programs for children with limb differences and provide research training opportunities for undergraduates in biomechanics and digital manufacturing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-09

This project aims to explore how the brain processes time when looking at images, with important implications for understanding memory, learning, and visual perception. The investigators have discovered that memorable images - those more likely to be remembered later - are also perceived as lasting longer in time than forgettable ones. This finding suggests that the perception of time relates to the gathering and remembering of visual information. The research team will conduct experiments using brain imaging, eye tracking, and computer modeling to understand how this process works across different parts of the visual system. This work could lead to better treatments for conditions like Alzheimer's disease, autism, and schizophrenia, where both time perception and visual memory are disrupted. The project will also advance artificial intelligence by creating new computer models that process images more like humans do, incorporating time as a feature that current AI systems ignore. The researchers intend to develop a public database of images with time perception information that can train AI systems to predict how long visual scenes will be remembered by humans, potentially improving a wide range of AI tools including educational software that adapts to how students process visual information. Technically, the project is designed to investigate whether memorable images undergo prioritized processing that dynamically extends temporal windows to maximize information extraction. Using behavioral experiments, electroencephalography (EEG), functional magnetic resonance imaging (fMRI), eye tracking, and recurrent convolutional neural network models, the project intends to pursue three main objectives: determining the extent of memorability's influence on time perception across different image features, mapping the neural mechanisms underlying this interaction throughout the visual hierarchy, and testing whether manipulating time perception can causally alter memorability. The experimental approach includes temporal bisection and reproduction tasks using images from established memorability databases, combined with multimodal neuroimaging to track spatiotemporal dynamics from early visual areas to higher-level timing regions. The project aims to use using recurrent neural networks to predict both memorability and perceived duration, testing the hypothesis that faster processing speeds in neural representations correspond to time dilation effects. The findings informs the development of next-generation AI systems that incorporate temporal processing as a core feature for processing visual information, addressing a critical gap in current computer vision models. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-09

The aim of this project is to extend the theory of calculus to more complex geometric settings, particularly those relevant to theoretical physics. Traditionally, calculus is defined on flat spaces like lines or planes, while modern differential geometry expands these concepts to smoothly curved spaces—such as spheres or donuts—and their higher-dimensional analogues, called manifolds. This theory plays a central role in physics, from Einstein’s description of gravity as the curvature of spacetime to the Standard Model of particle physics. The algebraic process of solving equations corresponds geometrically to the intersection of graphs, a principle that extends naturally to manifolds. However, intersections of manifolds are not always manifolds themselves, rendering differential geometry insufficient. The PI has made significant contributions to derived differential geometry (DDG), an advanced framework designed to handle such non-smooth intersections. Yet, integration—a cornerstone of calculus—has not been fully developed in this setting. The first aim of the project is to fill in this gap by building a robust theory of integration in DDG, with particular relevance to the computation of Feynman path integrals in physics. The second aim is to generalize geometric quantization—a powerful method traditionally used to describe the transition from classical mechanics to quantum mechanics—to more sophisticated systems such as classical field theories. Classical mechanics describes the motion of point particles, while field theories govern the behavior of extended objects, such as electromagnetic fields, and arise from purely mathematical constructions. Concrete outcomes of this project will include the development of new mathematical formalisms for integration and quantization over derived stacks, which can be used for computations in quantum field theory, such as path integrals and quantum invariants arising from topological field theories. These tools are expected to be applicable in both physics and mathematics, and the project will also foster interdisciplinary education by supporting the design of joint coursework in geometry, topology, and field theory for students in both disciplines. The project consists of two complementary components. The first involves constructing a comprehensive theory of geometric integration applicable to quasi-smooth derived higher stacks within derived differential supergeometry. This will be accomplished by developing a six-functor formalism, identifying dualizing complexes as Berizinians of cotangent complexes, and defining integration through the co-unit of exceptional inverse and direct image functors. The second aim of this project is to develop a notion of geometric quantization for shifted symplectic derived smooth stacks whose output is a fully extended topological field theory with values in higher categorical vector spaces. The resulting program will be a refinement of the shifted geometric quantization program of Safranov, appropriately adapted to the smooth setting, and will build on work of Calaque-Haugseng-Scheimbauer on TQFTs. The functorial field theories constructed using this method should yield new quantum invariants. By the cobordism hypothesis, such a theory is determined by what it assigns the point, and this should correspond to the underlying higher vector space of polarized sections of a higher prequantum line bundle. The PI proposes to prove that the fully extended framed TQFT associated to Chern-Simons theory is determined by a certain linear 2-category of representations of the string 2-group. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2025-08

Summary/Abstract We want to understand inter-individual differences in the clinical trajectory of chronic musculoskeletal pain (cMSKP) and the underlying factors that determine these differences. The research question we want to answer is the following: Can a time-varying network of multidomain symptoms be used to identify clinically-relevant transitions in pain and functional state, and do unique patterns of psychosocial, biomechanical and physiological factors determine these changes? Current management of cMSKP is focused on treating specific diagnoses or/and regional symptoms. There is a lack of diagnostic tools that capture interactions of symptoms across biopsychosocial domains and little guidance about how to interpret the relative contributors to the pain state and functional level. We will use knee osteoarthritis as a case study to test our hypothesis, although our findings can be generalizable for all cMSKP conditions. We will collect data from a prospective observational study of two cohorts recruited from community-based physical therapy (PT) clinics: those managing their knee pain with PT, and those undergoing PT after total knee arthroplasty (TKA). We will utilize a novel approach to integrate clinical measures with patient’s own knowledge and insights about their lived experience of pain and function through the following Aims: Aim 1: Develop network representations of patient’s lived experience of pain using large language models to annotate patient narratives. Aim 2: Develop network representations of patient’s functional state utilizing home-based videos of sit to stand functional tasks. Aim 3: Understand state transitions utilizing time series of daily pain, function, and psychosocial variables, and participant’s views of factors influencing pain and function. Our approach using multidomain network models represents a paradigm change to analyze and interpret data in cMSKP using a complex systems perspective. Our long-term objective is to develop clinically-feasible decision tools that enable whole-person assessment and management of cMSKP and guide personalized primary, secondary and tertiary prevention strategies. Our interdisciplinary team of physicians, physical therapists, engineers, neuroscientists and data scientists is uniquely qualified to tackle this challenging project.

NSF Awards · FY 2025 · 2025-08

The Paleocene-Eocene Thermal Maximum (PETM, 56 million years ago) is an important paleoclimate event used to understand how Earth’s climate system responded to rapid increases in atmospheric methane and carbon dioxide. One approach to study this interval is chemical analysis of microfossils known as foraminifera preserved in deep-sea sediment. Foraminifera grow calcium carbonate shells that record the environmental conditions of the organism’s habitat. However, PETM foraminiferal records suffer from two well-known limitations: first, ocean acidification at the PETM onset can lead to the dissolution of the carbonate shells and, second, vertical sediment mixing (bioturbation) can intermingle shells from different time periods together. Until recently, vertical mixing was a significant drawback because analyses required multiple shells to have sufficient accuracy. This project will remedy those issues by constructing records of individual foraminiferal geochemistry and morphology across the PETM at International Ocean Discovery Program (IODP) Site U1580 located on the Agulhas Plateau in the Southern Ocean. Site U1580 features abundant microfossils that were not dissolved. Cutting-edge analytical techniques will permit measurements on individual foraminifera to disentangle signals affected by bioturbation. Results will produce new estimates of surface and deep-water warming and carbon cycle dynamics across the PETM onset. The proposed work will support a team of three early career researchers plus graduate students and postdoctoral scholars. The project integrates educational outreach; investigators will create an open educational resource on the PETM and its relevance to contemporary climate. The presence of well-preserved foraminifera throughout the PETM onset at IODP Site U1580 offers a unique opportunity to reconstruct the magnitude, pace, and dynamics of climate change during the earliest phases of the PETM. However, pilot data demonstrate extensive vertical mixing of individual foraminifers across the event (as observed at other sites). The investigators will disentangle vertical sediment mixing by applying a series of measurements performed on individual shells. Shells will first be imaged by Scanning Electron Microscopy (SEM) and Computed Tomography (MicroCT) to characterize preservation and document morphology. Shells will then be analyzed for Mg/Ca (a paleotemperature proxy) via Laser Ablation Inductively Coupled Plasma Mass Spectrometry, then finally analyzed for their stable carbon and oxygen isotopic composition using a new CryoFocusing technique adapted for small carbonate samples. The resulting individual foraminifera carbon isotope data will distinguish pre-PETM from PETM individuals, allowing quantification of shell morphology, Mg/Ca-based temperature, hydrologic change, and the carbon isotope shift across the PETM onset. Any observed structure in the geochemical data will provide insight into the pace of change or lead-lag relationships between aspects of the carbon cycle and climate. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2025-08

PROJECT SUMMARY Loss of an upper limb can greatly impact the ability to perform daily tasks, work, and participate in social contexts, so a major goal of rehabilitation for individuals with upper limb loss is to facilitate their full engagement in all aspects of life. While a common research objective is to support these individuals through the development of more advanced prosthetic limbs, the experience of disability is also shaped by physical, biological, and social factors beyond the prosthesis itself. A human-centered approach to rehabilitation that addresses these interconnected factors could be more effective for enabling functional success in this population. An emerging concept that may be useful in this application is embodiment, or the perception that the prosthesis is part of the user’s body, but the relative importance of embodiment compared to other factors that influence prosthesis use is not well-understood. In particular, it is unknown if prosthesis embodiment is actually a desired outcome among individuals with upper limb loss. It is also unclear whether all individuals are equally capable of embodying their prosthesis, or whether they differ in their ability to acquire the requisite skills that would facilitate embodiment. There are numerous sources of error during prosthesis use, which come from both the user and the prosthesis, that could inhibit the experience of embodiment if individuals are unable to adequately compensate. Thus, variance in the experience of embodiment between individuals may be related to how they detect and adapt to with these error sources. The goal of this proposal is to better understand variation in the desire and ability of individuals with upper limb loss to experience prosthesis embodiment. In Aim 1, we will explore how individuals with upper limb loss perceive the necessity of prosthesis embodiment. We will conduct semi-structured interviews related to their experience or lack of experience with embodiment, using questions developed in consultation with individuals with upper limb loss. In Aim 2, we will examine the role of user-related and prosthesis-related error in determining movement adaptation and embodiment in upper limb prosthesis users. This will be accomplished using virtual cursor control tasks to separate the error sources and quantify embodiment, along with correlations to assess whether individual skill in error compensation is associated with stronger embodiment. Successful completion of this proposal will provide insight on the conditions under which upper limb prosthesis embodiment arises and the extent to which embodiment is necessary for optimal engagement in different life activities. This information can direct future assessments of prosthesis use to focus on the most impactful contributors to functional success. Ultimately, these assessments could support the development of prosthesis design and training protocols tailored to individual needs and preferences.

NIH Research Projects · FY 2026 · 2025-08

Uncoordinated eye rotation leads to binocular misalignment, such as strabismus, which affects over 18 million Americans. Strabismus is commonly treated surgically, but with suboptimal outcomes, with success rates reported from 30% - 80%. One cause of unsatisfied strabismus management is insufficient understanding of functions and synergetic control of the eye muscles. Engineering tools from imaging to modeling have been used to advance our understanding of eye neuro-biomechanics in normal and defective conditions. Physical systems that mimic human eyes have unique characteristics to obtain new knowledge on coordinated eye movement and its control. In this project, we propose to develop novel artificial muscle-driven robotic eyes that can realistically mimic binocular eye movement to study the biomechanics of human eyes and better understand the causes of strabismus and other eye movement disorders. The proposed study addresses three main questions to advance the fundamental knowledge gaps in modeling and control of oculomotor system: (1) how to model the highly nonlinear and complicated dynamic system in a simpler way; (2) how to create and control such a physical system with fast response and high precision; (3) how a human controls the oculomotor system, using what objectives. To address these questions, we will: (a) establish the computational model based on geometric algebra to describe the nonlinear dynamics of eye and compute the mystery moving pulley positions. (b) conduct tests on various types of artificial muscles to determine the most suitable material for mimicking eye muscles and optimize the fabrication process of producing artificial muscles that are close to eye muscles' anatomical and biomechanical properties. Robotic eyes actuated by artificial muscles that closely mimic the configuration of human eyes will also be developed. (c) To gain insight of underlying control strategies humans employ to control their eyes, we’ll utilize reinforcement learning with expert knowledge and real human/patient eye movement data to determine the objective functions humans use when controlling eyes. (d) Develop a patient-specific data-driven treatment planner for better strabismus management. This project aligns with NIBIB’s mission by developing an AI-driven, artificial muscle-powered robotic eye system to improve strabismus diagnosis and treatment. It will provide a novel platform for deeper understanding of ocular biomechanics, and enabling personalized, data-driven interventions, ultimately reducing reoperations and enhancing vision outcomes for millions.

NSF Awards · FY 2025 · 2025-08

The objective of this project is to support research on mathematical conceptualization for planning and making infrastructure investment decisions based on real options and risk tradeoff. Protecting civil infrastructures against natural hazards is crucial to the welfare of communities. Such protections involve major capital expenditure and often take years to complete, but not taking actions comes with detrimental consequences in the forms of economic loss, physical damage, casualties, and population displacement. There are difficult tradeoffs between taking early action when uncertainty is high and deferring them to a later point in time when uncertainty is low. While popular in finance, options theory has been utilized at only a basic level in civil engineering applications, typically focusing on a single element, ignoring system effects and uncertainty sources, risk tradeoffs, and assuming risk neutrality. This project seeks to contribute to fundamentally changing the framing of the underlying protective investment decision problem and advancing needed mathematics and algorithms to achieve this objective. The project involves four key research thrusts designed around: (1) advancing real option analysis methods under multiple uncertainty sources; (2) risk trade-offs and risk behavior modeling; (3) network-wide impacts and interacting options; and (4) machine learning approaches for information-rich environments. This research looks to build on concepts from optimal stopping theory, the Black-Scholes model, chance-constrained stochastic integer programming via p-efficiency, approximate dynamic programming, partially observable Markov decision processes, deep reinforcement learning, multi-agent extensions, and more. “Learn-to” algorithmic approaches that embed machine learning within optimization algorithms for speed-up, and explainable artificial intelligence for actionable protocols plan to be developed. Educational and outreach activities include educational modules/mini-videos; presentations to decision-makers; input from local decision makers; involvement of undergraduate students and cross-university events and courses. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2025-08

Abstract The overarching goal of this five-year grant application is to develop a technological toolkit that enables noninvasive mapping of ionic activity in the brain of a mouse. The proposed in vivo imaging toolkit relies on an innovative voltage-sensing nanoprobe comprising of indocyanine green (ICG) dye, and a quencher, templated on DNA nanoparticles (DNA-NPs), which offers several advantages: 1) 805-nm excitation and fluorescence emission that extends to near-infrared-II (NIR-II, 1000-1700 nm): low background signal and deep tissue (centimeter) penetration due to low autofluorescence and suppressed light scattering at these wavelengths; 2) Precision nano-engineering: pre-determined organization of dye molecules for high local density and increased photostability; and 3) Cell-specific labeling: DNA-NPs can be conjugated with peptides, aptamers, or antibodies to enable cell-specific targeting. We are working closely with Photon etc, a leading industry entity that won the 2019 World Molecular Imaging Society Commercial Innovation award for developing IR VIVO™ NIR II imaging system, to establish the utility of our DNA-based voltage reporting probes for NIR-II mapping. The IR VIVO™ system can offer multi-scale imaging with 20-µm resolution in 5 mm field of view to coarser resolution with a wide lens at fields of view capable of imaging the whole animal. We will first modify our already working probes to selectively stain the inside or outside of cells as varying distances to probe ionic imbalances induced by neuronal activity. We will demonstrate cell-specific labeling of NIR-II potassium imaging in vitro using cultures of neurons, as well as ex vivo hippocampal preparations. Then, we will image ionic activity in the mouse brain through the intact skull and skin by observing the well- established acute pilocarpine seizures. These in vivo imaging studies will demonstrate the potential of the proposed NIR-II imaging toolkit to noninvasively map ionic activity in the brain. These tools will enable new preclinical investigations such as (1) examining the role of ionic dynamics and imbalance in neurodegenerative diseases (2) understanding how a breakdown in ionic balance can be used to identify seizure foci and (3) guiding and monitoring electroceuticals (aka bioelectronic medicine) that can be used to treat or lesion pathological tissue. Furthermore, our ion-reporting probes are made of DNA and FDA-approved ICG dye. Therefore, these biocompatible probes can also be used for mapping bioelectrical activity in human subjects for translational research and clinical applications, e.g., image guidance for minimal tissue ablation during surgical procedures to treat intractable epilepsy or to assess functional recovery after traumatic brain injury.

NSF Awards · FY 2025 · 2025-08

This project envisions a prosperous and secure Arctic region focusing on Alaska that can build, maintain, and operate resilient and sustainable coastal and interior civil infrastructure and can adapt to the dynamic marine and terrestrial environmental changes. This vision will be achieved by engaging with Alaskan communities, industry, and local-to-federal government entities, thereby building a pipeline for workforce development of future scientists, engineers, and skilled workers with expertise in Arctic environments. The team will collaborate with the North Slope Borough and the communities in Seward Peninsula to co-develop and implement the solutions to emerging challenges, notably coastal and riverine erosion in the Arctic coastal communities, infrastructure failures induced by permafrost degradation, and flooding. The resilience solutions and technologies, from ideation to implementation, will be co-developed through close collaborations with partners of Indigenous communities, industry, local to federal government, and six academic institutions. The impacts include improved well-being and resilience of individuals and communities in the U.S. Arctic, increased economic competitiveness of the U.S., improved national security, and increased public scientific literacy and public engagement with science and technology. The project will generate new understanding of how the Earth system (including the northern and northwestern Alaska region, permafrost, and coast-land interface) changes, and its interactions with the built and sociocultural systems, thus building the foundational knowledge base to develop solutions to emerging problems. At the end of Phase-1, the project will (1) identify and specify the solutions needed to address the U.S. Arctic challenges from permafrost degradation, erosion, and flooding, (2) identify data gaps and devise approaches to collect new data for the technology development, (3) define specific requirements for the technologies and solutions, and (4) identify application sites for the technologies and solutions and collaborating partners. Project costs and feasibility in translation of research to solutions will be demonstrated by conducting techno-economic analysis on enabling technologies and system-level solutions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

This collaborative project explores immersive video streaming technology that lets people see and interact with lifelike 3D environments. Instead of watching a regular video on a flat screen, users can move around a scene, change their viewpoint, and experience the environment from different angles in real time. This is made possible through advanced techniques that simulate how light travels through a scene using artificial intelligence. By leveraging computer vision, computer graphics, and machine learning, the project aims to make these 3D experiences smooth, detailed, and fast, even over current Internet connections. This project builds a high-performance immersive video streaming system using neural radiance fields, an advanced 3D scene representation powered by machine learning. It is structured around three key thrusts: improving resiliency, optimizing bandwidth, and reducing latency. Resiliency is addressed through lightweight models that recover lost content, adaptive encoding based on content robustness, and artifact reduction via reprojection. Bandwidth is optimized by filtering less important content, prioritizing semantically significant regions, and applying hybrid upsampling. Latency is reduced through intelligent packet ordering, caching intermediate results, and collaborative edge-client rendering. These innovations are integrated into a unified framework called NeuVol, which will be evaluated on diverse hardware and real-world network conditions. The project’s broader impacts include both technological innovation and societal advancement. It will enhance immersive applications in fields such as remote training, collaborative design, and medical diagnostics by enabling efficient delivery of high-quality 3D content. Educationally, it will integrate research outcomes into courses on multimedia systems and immersive computing, offering students hands-on experience with emerging technologies. The project will actively engage undergraduate researchers through active mentorship. Public outreach will feature demonstrations at schools, community centers, and science fairs to promote awareness of immersive technologies and inspire interest in engineering and computer science among the broader public. The project brings together researchers from George Mason University and the University of Illinois Urbana-Champaign to design, build, and test NeuVol and make it available to a wide audience through open research. All research artifacts developed through this project will be made publicly available via a dedicated project website at https://neuvol.github.io. The website will be actively maintained throughout the project duration and archived for continued access beyond the project’s conclusion. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

From the hierarchy of search engine results that dictate the ease with which we find information, to the curated news and social media that we view, and the prioritized online contacts forming our social and professional networks, learning to rank (LTR) is an indispensable mechanism. Beyond meeting routine information needs of general users, LTR is also essential in supporting specialized decision makings, such as filtering healthcare documents for diagnostics or prioritizing legal documents in legal practices. Hence, ensuring that these rankings are trustworthy, by making the reasoning process transparent, ensuring ranking distribution balanced across different users and items, and maintaining robustness in unpredictable real-world conditions, is of paramount importance. Current approaches, however, often address superficial symptoms while overlooking fundamental causes, leading to limited effectiveness, compromised transparency, and thus eroded usability and credibility. The objective of this project is to leverage neuro-symbolic learning to develop novel ranking algorithms with inherently interpretable reasoning processes. The integration of expressiveness in deep learning and transparency in symbolic reasoning presents a compelling avenue to synergistically capitalize on the merits of both realms. Moreover, owing to the innate transparency of neuro-symbolic ranking models, these models readily identify factors within the reasoning process that lead to various prediction imbalance and safety issues. Uncovering and investigating these elements enlightens more effective, holistic, and transparent approaches toward balanced and robust ranking systems. This project aims to forge novel neuro-symbolic learning to rank models to deliver interpretable, balanced, and robust rankings through four thrusts. The first thrust introduces a logic neural network that transparently elucidates the entire inference process in ranking and presents a logic AutoEncoder, a type of artificial neural network for efficient coding, for interpretable collaborative filtering rankings. Moreover, a neuro-symbolic representation learning technique is proposed to enhance model expressiveness and accuracy. Second, this project proposes to utilize the inherent transparency in neuro-symbolic ranking algorithms as key tools, to effectively enhance ranking balance for both users and items. Last, this project proposes novel strategies leveraging and augmenting the interpretable reasoning process of neuro-symbolic learning to fortify robustness against shortcut features, exposure differences, and data poisoning attacks. This project will also apply proposed research innovations in two real ranking systems, organ transplantation management and research paper recommendation, to demonstrate the effectiveness of proposed technologies in real-world contexts and to ensure the proposed project would be successful with impact. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

Modern transportation systems have undergone a significant transformation, marked by increased design complexity, advanced networking capabilities, and an overwhelming surge in data. As a result, today's automotive system is a collection of interconnected embedded systems, some of which (such as the infotainment system) are also connected to the Internet. As the number of connected vehicles grows, traffic systems become networked and autonomous fleets emerge in the consumer space, the potential for cyberattacks on U.S. transportation infrastructure increases significantly. Given the criticality of the transportation cyberinfrastructure (CI), this project builds expertise in the automotive cyber domain through development of testbed and training curriculum material and summer training workshops for educators, students, and researchers. The project addresses critical issues in cyber workforce development in the transportation and automotive sectors through three initiatives. The first leverages faculty from different disciplines to develop a coherent open-source CI that provides a unified research platform for automotive and autonomous systems. Leveraging this CI, the second initiative delivers modular training materials on secure transportation system design. Finally, the third initiative arranges a series of workshops to directly train 80 participants, including faculty members, graduate students, and cyberinfrastructure professionals. The project also produces and disseminates royalty-free resources to support workforce development and education in transportation system security. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

This award provides support to 25 U.S.-based graduate students to attend the 2025 Conference on Computer and Communications Security (CCS), to be held in October 2025. CCS is the flagship conference of the Special Interest Group on Security, Audit and Control of the Association for Computing Machinery (ACM). The conference brings together information security researchers, practitioners, developers, and users from all over the world to explore cutting-edge cybersecurity ideas and results. CCS 2025 will include multiple technical tracks across a wide spectrum of cybersecurity topics, with hundreds of papers, pre-conference and post-conference workshops, tutorial and poster sessions, and panel discussions. Participation in the conference is a valuable opportunity for students and is an important part of graduate education in cybersecurity. The conference has special events aimed at connecting student attendees with mentors, further increasing the benefits students receive. Students will have the opportunity to observe high-quality presentations and interact with senior researchers in the field both in the main conference and the associated workshops. This can lead to community-based research initiatives, knowledge sharing, and positive social impacts beyond academia. The conference will involve researchers, scholars, and professionals from diverse backgrounds, allowing students to build valuable connections and collaborations. To broaden perspectives and participation in the discipline, the organizing committee will conduct a wide outreach program to solicit applications, including from students in underrepresented groups who might not otherwise be able to attend. Criteria for selection include evidence of a serious interest in the field as demonstrated by the applicant's description of their research background and plans, as well as their ability to both benefit from and contribute to the conference. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-07

Harmonic analysis is a major branch of mathematical analysis which focuses on studying the behavior of functions by breaking them down into simpler and easier-to-understand component parts. The developments in harmonic analysis have led to concrete advances in medical imaging, image compression algorithms, signal processing, and neuroscience. This project examines questions in harmonic analysis and related fields from a more theoretical or pure perspective of basic research. As part of this award, the PI also mentors undergraduate students in research projects, which increases the STEM pipeline and supports higher education and society at large. This project consists of two main streams: harmonic analysis in the special setting of non-doubling measures and applying harmonic analysis to problems in complex analysis which also connect to operator theory. In the context of non-homogeneous harmonic analysis, questions relating to a novel paradigm for the sparse domination of Calderon-Zygmund operators and commutators are considered. Within the second stream, the PI investigates two-weight and endpoint commutator estimates for the Bergman projection, Lp estimates for the Cauchy-Szego and Bergman projections on Lipschitz and other minimally smooth domains, and two-weight inequalities for the Bergman projection on the unit disk. The specific tools used to study these questions include a non-homogeneous Calderon-Zygmund decomposition, an approach to the study of holomorphic projection operators originated by Kerzman and Stein, and using weighted Haar decompositions and random dyadic grids to achieve two-weight estimates. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

Deep Learning (DL) has improved scientific applications across various scientific domains, including high-energy physics, meteorology, agriculture, and material science. This project introduces DLToolkit, a performance profiling infrastructure tailored for domain scientists to analyze and optimize science-driven DL applications. This project also contributes to education and supports broader usage; the outcomes of this project will be integrated into the Computer Science (CS) curriculum, and both George Mason University and the University of California - Merced are minority-serving institutions, offering opportunities for delivering knowledge about cutting-edge techniques to underrepresented students. Together with industry and national laboratory partners, the project will also provide research training, symposia, and internship opportunities for students, aiming to foster a cohort of performance engineers. The overarching objective of this project is to improve scientific DL applications. The intellectual merits include three novel profiling capabilities: (a) synergistic tool-framework profiling to streamline extensive domain-specific knowledge from existing DL frameworks to DLToolkit, significantly lowering the barrier for domain scientists to use DLToolkit; (b) just-in-time (JIT)-aware profiling to ensure precise yet lightweight attribution of performance events to complex JIT-compiled DL operators; and (c) tensor-centric profiling to provide a holistic view of tensor operations’ impact on model performance. By uniting these capabilities within DLToolkit, this project will create a cohesive infrastructure for domain-specific performance profiling to empower scientists with critical insights to optimize their DL applications, accelerating scientific research and innovation. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

Despite recent impressive progress in the creation of language technologies, thousands of languages spoken by millions of humans are not yet unsupported by modern technologies like large language models, translation tools, or chatbots. Typically, the bottleneck towards building such tools is the lack of data for these languages. This project proposes to instead bypass the data-scarcity issue by using grammars, effectively leveraging the already-codified linguistic knowledge about the languages themselves. In doing so, this project will provide a concrete path towards building multimodal language technologies for data-scarce languages. Concretely, this project will (1) build the necessary datasets to explore the meta-linguistic capabilities of large language models. Then, it will (2) explore ways to build LLMs for new languages by relying on grammars, similar to how a human learns a second language. Next, the proposed work aims at (3) making theoretical connections to various learning paradigms, and attempting to model the process of multilingual learning itself. Last, the project will (4) delve deeper into the errors potentially exhibited by language models. Additionally, the proposed research will be integrated into education through new teaching modules combining linguistics with natural language processing, promoting undergraduate research. The project will aim to extend its impact beyond the classroom through close collaboration with underserved language communities, aiming to build technologies according to the communities' needs. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

Along with the emergence of actual quantum computers, quantum computing is expected to transform science, technology, and society by solving problems intractable for classical computers, including those in chemistry, materials, and finance. However, existing quantum computers face a major obstacle: their results are often unstable over time and difficult to reproduce; this is primarily because of the fluctuating noise in the environment. The stability challenge exists across different qubit technologies, including superconducting qubits, trapped-ion qubits, and neutral atom qubits. To address this pressing challenge, the Quantum System Stability and Reproducibility Workshop (StableQ) was established at the Quantum Week 2023, a conference sponsored by the Institute of Electrical and Electronics Engineers (IEEE), followed by its second edition at the IEEE/ACM International Symposium on Microarchitecture (MICRO) in 2024. The third edition, StableQ 2025, is scheduled to be held in September 2025 at IEEE Quantum Week at the Albuquerque Convention Center in New Mexico. This award will support students and researchers to participate in StableQ 2025. The project will bring together experts from academia, industry, and national laboratories to report state-of-the-art developments, exchange ideas and practices, and foster cross-disciplinary research aimed at addressing unstable noise and enabling reproducible quantum computing. The workshop is uniquely positioned to foster collaborative efforts by bringing together experts from hardware, software, and algorithms to tackle key stability issues. Through interactive sessions, participants will identify challenges, prioritize research directions, and address gaps between academia, industry, and national laboratories. Ultimately, this project aims to develop novel techniques, strengthen collaborations in the research community, and deliver scalable, stable, and reproducible quantum computing platforms. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2025-06

Abstract In the last three decades, extensive research has been conducted on the effects of prenatal alcohol intake on birth outcomes, fetal alcohol spectrum disorders, and children’s developmental delays. However, there is limited research on highly prevalent comorbidities that exist among pregnant women who consume alcohol during pregnancy, and how these maternal comorbidities in conjunction with prenatal alcohol consumption affect birth outcomes. Given the high national rates of obesity and subsequent diabetes, pregnancy-induced hypertension, and related preeclampsia and toxemia along with persistently high rates of preconceptual and prenatal alcohol consumption among US women, this gap in our science is significant. The proposed study will address this gap using an innovative approach that yields risk profiles that may be easily translated into clinical screening tools and provide the underpinnings of tailored interventions for the target population. The objectives of this study are to: (1a) estimate the extent of comorbidities among pregnant women with prenatal alcohol exposure; (1b) build maternal morbidity risk profiles based on social and behavioral determinants of health among women with prenatal alcohol intake using machine learning (ML) methods; (2a) examine the effects of comorbidities on adverse birth outcomes among alcohol-exposed pregnancies; and 2(b) build risk profiles for adverse birth outcomes using ML methods. We will use data from the Prenatal Alcohol and SIDS and Stillbirth Network (PASS) (2007-2015) and the Pregnancy Risk Assessment Monitoring System (2000-2021). Logistic regression will be used to compute odds ratios and 95% confidence intervals in examining the direction and magnitude of the association between comorbidities and alcohol exposure during pregnancy. Further, we will examine the effects of comorbidities on adverse birth outcomes among alcohol- exposed pregnancies using traditional statistical methods. ML methods such as classification trees will be used to construct risk profiles for maternal morbidity and adverse birth outcomes in identifying high-risk groups based on exposure to prenatal alcohol intake. This study will provide vital information on comorbidities among women who consume alcohol during pregnancy. The proposed study presents a model of study that combines both maternal and infant health in the context of maternal-fetal exposure to alcohol using innovative methods of data analysis that will yield previously unidentified risk profiles. Given the need for careful allocation of scarce healthcare resources for prevention and treatment programs, the identification of these risk profiles is both innovative and critical.

NSF Awards · FY 2025 · 2025-06

This project explores how scientific messages about environmental adaptation shape people’s decisions, especially when their beliefs about the environment, politics, and other issues are closely connected. By examining why some individuals embrace adaptation measures—such as flood-proofing homes or relocating from high-risk areas—while others do not, the project seeks to clarify how these choices can unintentionally deepen social and economic inequalities. In doing so, the project addresses the national interest by improving public understanding of climate risks, fostering more informed decision-making, and promoting inclusive, effective adaptation strategies. The project also advances broader societal goals by training undergraduate researchers, partnering with nonprofit organizations to develop practical outreach methods, and creating software tools that can be used to study belief systems in various contexts. This project develops a new conceptual framework, new methodological tools, and survey-based measurement approach for “belief networks,” capturing how an individual’s attitudes about topics such as climate, people migration, and social fairness simultaneously influence one another. Large-scale surveys in the United States and a case study in Virginia will map how these networks vary across partisan, demographic, and socioeconomic groups. The researchers will then design and implement scientific communication experiments to examine how information targeting specific attitudes—and the links among them—affects people’s adaptive capacity and motivation. Alongside these empirical studies, the project supports new network science methods to analyze individual-level belief systems at scale, providing computational and statistical tools for extracting, modeling, and comparing complex attitude linkages. By integrating these methodological advances with science communication research, the project offers insights into how scientific information can most effectively encourage robust, evidence-based adaptation decisions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

As computing infrastructure gradually transitions from classical high-performance computing to quantum-centric computing cyberinfrastructure (QuCI) with quantum computers, classical computers, and AI accelerators, QuCI is expected to revolutionize domain applications such as computational chemistry, material science, and combinatorial optimizations, outperforming state-of-the-art classical computing significantly regarding speed, accuracy, or the ability to handle larger problem sizes. This project proposes an automated QuCI deployment framework (AutoQC) to perform quantum application deployment automatically. Beyond the new technology, the project also supports a cohort of education and outreach activities, including hosting workshop series at regional minority-serving institutions, organizing events at interdisciplinary communities (e.g., workshops, tutorials, competitions), initializing K-12 programs, and developing and updating curriculum. One key component in these activities is developing a novel visualization tool that makes the decision-making process visible. It also serves as an educational tool for curriculum and outreach activities to lower the learning bar for beginners. The project designs automated, efficient, scalable AutoQC algorithms across full-stack deployment layers, including resource allocation, circuit compilation, and functional circuit design layers. Specifically, this project includes (1) designing a high-quality AI-powered quantum performance predictor, yielding an innovative fidelity-aware QuCI resource allocation tool; (2) developing an application-specific pulse-level compilation algorithm to improve system stability at run-time on noisy quantum devices; and (3) building an automated quantum circuit search technique to construct logical quantum gates for a specific quantum error correction code. This research agenda enables the automated deployment of a given quantum circuit to physical quantum bits with improved usability, efficiency, scalability, and reliability on QuCIs at both near-term noisy intermediate-scale quantum computing and long-term fault-tolerant quantum computing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

This I-Corps project focuses on the development of secure and energy-efficient software update mechanisms for large-scale, industrial, Internet of Things (IoT) networks. Industrial IoT networks have specific and unique operating requirements that make them highly prone to service disruption and security breaches. Moreover, these networks typically cover large geographical regions and operate in mission-critical industries such as healthcare and energy production. These challenges make software update processes highly challenging, and a top requirement by industrial IoT service providers. This solution also has the potential to decrease the threat of security breaches and attacks on mission-critical infrastructure in the energy, healthcare, and military sectors. As a result, this technology could reduce the economic costs incurred by service disruptions and any accompanying reputation loss, which can reach several millions of U.S. dollars per security incident. This I-Corps project utilizes experiential learning coupled with a first-hand investigation of the industry ecosystem to assess the translation potential of the technology. The solution is based on the development of energy-efficient cryptographic techniques to ensure the incremental updating of sensor-nodes in large-scale, industrial, Internet of Things (IoT) architectures. The technology safeguards the confidentiality, integrity, and authenticity of the software update components, and minimizes the service disruption time needed while loading, deploying, and processing the software updates. The main security engine leverages probabilistic data structures and incremental cryptographic mechanisms to reduce the energy consumption of the software update algorithms and to ensure the uninterrupted operation of the industrial IoT network during the update process. This secure software update solution could provide a safer operating environment for end users in applications relying on IoT architectures. Moreover, the energy-efficient and incremental cryptographic data structures employed in this solution increases the scalability of the overall IoT network and reduces service disruption times, lowering the overall operating costs of mission-critical industrial IoT deployments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-06

The objective of this project is to support effort in mobilizing a collaboration of researchers and practitioners to create innovative, minimally disruptive and potentially transformative solutions to how we design, build and operate urban underground infrastructure systems. Cities in the United States and across the world have been a major driver of economic growth, technological innovation, and cultural vitality. However, their infrastructure systems are often patchworks of legacy and new components with incompatible standards, materials, and governance structures. As a result, the performance of such systems can be unpredictable under normal conditions and more so when subject to extreme events. The current economic paradigm requires seamless and continuous service delivery, supporting uninterrupted movement and commerce. The challenges to delivering such services that are faced by underground infrastructures, such as water and wastewater, transportation, telecommunications, and power systems, are exacerbated by difficulties in access and the harsh environment in which these systems reside. Key knowledge gaps remain a barrier. Through meetings, exchanges, workshops, resource sharing, training and more, the Transforming Urban Underground Infrastructure (TUUI) Collaborative provides a forum for information sharing, idea generation and opportunity creation. It fosters new collaborations and partnerships, develops a community of experts around urban underground infrastructure and advances science and a vision for the future in urban underground infrastructure systems. The collaborative brings together experts from across disciplines, technical specialties and geographical locations to advance fundamental knowledge in designing, constructing, operating, controlling, maintaining, protecting, and improving urban underground infrastructure systems. It fosters activities around advancing technologies, fundamental theories, designs, operations, methods and practices for urban underground infrastructure systems and the underground surroundings. Activities enable communications, coordinate teaming opportunities, and develop an online repository of training and educational materials and other resources. With its working group structure and open, expert-led avenues for participation, new research partnerships and paths for coordination are formed that aim at transforming urban underground infrastructure given recent breakthroughs in computing, sensing, machine learning, automation, materials, and more. This project promotes the progress of science in underground infrastructure, assists in advancing national prosperity and welfare, and bolsters knowledge that can support national defense. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-05

Organizations and individuals outsource sensitive data from their own local systems to remote cloud services, making encryption necessary to maintain the privacy and security of the sensitive data. However, to preserve the usefulness of the outsourced data, the goal is to encrypt it so that users can still ask questions about their data without ever decrypting it on the cloud's premises. Current end-to-end encrypted system approaches aimed at addressing this goal have converged to two extremes. In one case, the approaches leak (i.e., reveal) some information about the encrypted data records that are processed in answering a query, which gives a false sense of security since leaks may enable reconstruction of the underlying sensitive information by sophisticated adversaries. In the other extreme, approaches do not reveal any information about the processed data, which requires significantly more computation and makes the overall performance too slow for practical use. The project's novelties are introducing a new paradigm for encrypted systems with carefully constructed, fine-grained leakage that is (1) scalable for practical use and (2) provides rigorous guarantees of protection against the reconstruction adversaries. The project's broader significance and importance are in influencing practical designs and industry practices to create a safer deployment of encrypted systems. The project aims to enhance education by integrating security modules into the undergraduate and graduate curriculum, introducing interdisciplinary courses, broadening student engagement, and expanding pilot programs with local high schools. This project introduces encrypted systems with fine-grained leakage and focuses on three interconnected areas: The first explores how specific query and update patterns can nullify leakage, providing provable guarantees of inapproximability or inability to reconstruct against adversarial reconstructions. The second thrust develops cryptographic designs for structured encryption, enabling context-informed leakage profiles that dynamically increase adversarial uncertainty. The third thrust applies this paradigm to concrete real-world scenarios. The research team plans to share project results with industry collaborators. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.