Stevens Institute of Technology

NSF Awards · FY 2025 · 2025-08

This project aims to advance the efficiency of machine learning model inference by developing a compression-aware computing framework. As large machine learning models become increasingly essential across science and industry, their substantial resource demands, particularly memory, computation, and energy, have limited their deployment to expensive, specialized hardware. Lossy compression techniques, such as sparsification and quantization, have emerged as solutions for running these models on consumer devices, such as laptops and smartphones. However, these methods often reduce predictive accuracy and require extensive tuning. This project proposes an approach to address this accuracy-efficacy trade-off by building self-aware machine learning models over lossy compression. By enabling models to detect and adapt to their own compression, this project will unlock pathways for cost-effective machine learning model inference. Technically, the project consists of two core research objectives. The first focuses on developing machine learning models capable of self-awareness in response to lossy compression. This includes enabling models to determine whether they have been compressed, identify the type of compression used, and localize the affected components. The second objective leverages this self-awareness to recover and enhance model performance without retraining. Techniques include instruction-based recovery, sparse zeroth-order optimization that adjusts a small subset of model parameters, and a collaborative inference framework where multiple compressed models work together. The project will evaluate these methods on real-world tasks such as long-context language modeling and biomedical question answering. By addressing fundamental limitations in compressed machine learning model inference, this project will contribute practical tools for efficient machine learning model deployment. 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 award supports participants of the conference "Groups, Logic, and Computation: Interactions between Group Theory, Model Theory, and Computer Science (GAGTA 2025)" which will take place at the Stevens Institute of Technology (New Jersey), June 9--13, 2025. The event will bring together researchers from various branches of group theory, model theory, and computer science to explore open questions in the field, now being approached from fresh and promising perspectives. It will also strengthen the discipline's connections to other branches of mathematics. Through this exchange of ideas among experts, students, and postdoctoral researchers, the conference aims to disseminate current knowledge and identify promising directions for future research. The conference will focus on recent developments in group theory, emphasizing groups and group actions, as well as their applications across various areas of mathematics where they serve as fundamental tools. The program will cover multiple branches of modern group theory with a particular focus on geometric, asymptotic, and combinatorial group theory, dynamics of group actions, probabilistic and analytic methods, first-order rigidity and classification, and Diophantine problems in groups and rings. Additionally, the conference will explore emerging AI connections, with dedicated sessions examining how group theory can further impact machine learning, formal verification, and symbolic computation and how AI methods may contribute to advances in group-theoretic research. This interdisciplinary exchange aims to foster collaboration and open new directions in both mathematics and AI. More information can be found at https://web.stevens.edu/algebraic/Stevens2025/ 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

A “liquid bridge” is a small body of liquid connecting two solid surfaces that maintains its shape due to surface tension. When liquid bridges connect solid particles, they cause the particles to adhere to each other. For example, sand castles owe their stability to water bridges that connect sand particles. Liquid bridges can also cause plugging of oil pipelines by forcing aggregation of gas hydrate particles. This project will use numerical simulations to examine the dynamics of liquid bridges between particles. The fluid dynamics of the bridges will be analyzed when the particles are moved apart or oscillated. Results from the project will help improve predictions of the behavior of wet particulate materials. The project will also support training of students in advanced methods of numerical simulation. This project will investigate the dynamics of liquid bridges between particles using Lattice Boltzmann Method (LBM) numerical simulations. The LBM is a computational fluid dynamics (CFD) technique that is well-suited to multiphase flows with complex geometry. In many engineering situations, liquid bridges can form between freely-moving solid particles. These bridges can rupture due to particle motion or can force multiple particles to aggregate into macroscopic structures. Previous simulations intended for static or quasistatic situations cannot capture these dynamic processes. This project will customize LBM simulations to elucidate the rupture of liquid bridges undergoing extensional, shearing, or vibrational deformations. The simulations will capture the two-way coupling between particles and fluid, the effects of inertia and viscosity, the effects of particle shape, and the effects of contact line pinning, e.g. due to surface roughness or faceted particles. The predictive capability developed by this project will benefit multiple applications, especially plugging of oil pipelines due to hydrate particles held together by liquid bridges. However, other applications involving liquid bridges, e.g. the mechanics of wet soils, gravure printing, or 3D-printing of particulate materials will also benefit. 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 · 2025-01

Next generation (NextG) network systems are envisioned to be complex, ubiquitous, and smart, which are likely to consist of millions of heterogeneous mobile devices to connect everything digital, enable machine-to-machine communications, and support a variety of critical machine learning (ML) paradigms, including the most popular federated learning (FL) over mobile devices. However, stakeholders in many intelligent mobile applications/services are resource constrained in terms of spectrum, energy, computing, etc., which poses many challenges to FL inspired applications/services. This project targets to develop a novel NextG network with high degrees of resiliency to address those challenges, in particular, when there may be massive bursty workloads, insufficient spectrum availability, limited computational and storage capability on edge, and privacy concerns of the training data on mobile devices. The anticipated project outcomes will enrich the knowledge of wireless systems and machine learning technologies and provide multidisciplinary training especially for underrepresented students. Additionally, the findings and innovations will be shared across the 23-campus California State University (CSU) system, where 90% of campuses are minority-serving institutions. Outreach activities including high school internships and summer undergraduate training programs can provide early exposure to research in science and engineering, fostering interest and encouraging more female and minority students to pursue careers in these fields. This project aims to address the resilient issues of FL over mobile devices via a novel holistic NextG network design across network architecture, local mobile devices, and accessing networks. (1) From the networking system's perspective, to support FL over large-scale heterogeneous mobile devices, serverless computing is exploited at the edge to resiliently and efficiently provide ML computing as a service. (2) From the local mobile devices' perspective, to resiliently protect local training data privacy against inference attacks in FL, an energy-efficient piggyback differential privacy (DP) design is proposed by jointly considering DP amplification from gradient quantization and sparsification, and free Gaussian noises from wireless channels. (3) From the accessing networks' perspective, to improve the spectrum accessing resiliency, network scalability, and spectrum efficiency, a multi-bit over-the-air computation (M-AirComp) based spectrum accessing design is proposed, which can enable efficient transmission of FL model updates even with limited spectrum availability, reducing the total energy consumption for mobile devices. 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 · 2025-01

The swift advancement in new energy sources, industrial automation, automotive controls, and space exploration demands the investigation of harsh-environment electronics, often withstanding temperature ranges from hundreds to several thousands of degrees Celsius (°C). For instance, space science communities have advanced electronic components to explore Venus’s atmosphere, which can reach temperatures above 500 °C. Turbine engines in aerospace industries utilize high-temperature sensors for remote pressure transducer interfaces, digitally interconnected actuators, and digital engine controls. The 4th-generation (Generation IV) reactors will function at coolant temperatures higher than light water reactors, reaching around 1000 °C. Hence, the instrumentation and sensors utilized for real-time health monitoring must function in these harsh conditions. Lastly, hypersonic vehicles reach speeds greater than Mach 5, where aerodynamic heating affects air flow, resulting in temperatures exceeding 1000 °C on the vehicle’s surface. As a result, surrounding air molecules ionize and create a buildup of plasma that interferes with electromagnetic waves. Despite its importance, technical challenges exist in radio frequency sensing and communications associated with high-temperature environments. This project will establish new research fields in extreme environments, high-frequency sensing, and high-temperature communications. Because enabling sensors and communication systems in extreme conditions is of paramount importance in designing satellites, spacecraft, automobiles, and space-exploring scientific probes, the proposed methods will benefit future applications including massive global satellite communication networks realized by tens of thousands of Earth-orbiting nano-satellites, hypersonic delivery/transportation infrastructure, CubeSat-based planetary sensing, as well as many other possible commercial and industrial applications. This CAREER project will integrate research and education programs and provide excellent opportunities for high school and college undergraduate/graduate students to engage in STEM research. This CAREER project aims to establish foundational high-frequency electronics for extremely high-temperature sensing and communications beyond 1000 °C, targeting broad industrial applications as well as aerospace and defense applications affected by hypersonic radio blackout. As far back as the 1960s, aerospace communities conducted flight tests to examine radio interference. However, nearly 70 years after the beginning of space exploration, the hypersonic vehicle’s radio interference remains an unsolved problem. In recent years, antennas operating above interference cut-off frequencies and metamaterial-inspired structures have shown new directions, but they lack practical implementations incorporating realistic interference magnitude, effective sensing techniques, system integration and validation, and high-temperature effects. The proposed research rests on the premise that dielectric and ceramic materials exhibit excellent electromagnetic (EM) properties, thermal isolation, and heat tolerance. They are ideal for high-frequency radiating components and for integration with active circuits to build hypersonic transponders and sensors. The project will start with experimental modeling of ceramic materials’ EM properties and thermal expansion in high-temperature conditions and investigate innovative temperature-dependent compensation techniques across extensive temperature ranges, at microwave and millimeter-wave frequencies. The scope and approaches to overcome radio interference challenges include: (1) 3D printed all-ceramic meta-structure, probe, and high-temperature apparatus to compensate for extreme-heat-induced plasma layers, (2) frequency-agile high-temperature/plasma sensing spectroscopic reflectometers with digital compensation for probe’s temperature-dependent EM properties, (3) phase-noise tolerant retrodirective signal trackers with on-chip analog signal processing, (4) artificial-intelligence-assisted temperature-compensating ceramic fiber harnesses. This CAREER project will enable reliable RF signal transmission/reception and signal integrity monitoring with unprecedented dynamic range and sensitivity in continuously and rapidly changing high-temperature environments for hypersonic applications. This CAREER project will also offer new techniques to integrate thermal protection systems and RF electronics, simultaneously achieving thermal isolation and electromagnetic propagation for extremely high-temperature sensing and communications. 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 · 2025-01

The synergy of the Internet of Things (IoT) and artificial intelligence (AI), also known as the Artificial Internet of Things (AIoT), has been catalyzing numerous smart systems by moving AI closer to data sources, enabling low-latency or even real-time intelligent on-site decision-making. The synergy, however, also brings forth unprecedented challenges in data security and privacy due to the vulnerabilities exhibited in both AI and IoT. By addressing these critical cybersecurity challenges, the project seeks to enhance national resilience against cyber threats and contribute to the advancement of secure AIoT systems essential for modern society. This project investigates data and model attacks and defenses on AIoT devices and studies secure computation offloading schemes on AIoT devices through several research tasks. Toward addressing these challenges and extending sustainable research capabilities in Cybersecurity and AIoT of York College, this three-year project will establish long-term cross-department and cross-institution collaboration platforms, including a new Cybersecurity and AI lab at York, a summer visiting student research program at Stevens Institute of Technology and a joint AIoT research testbed. This project centers around data security and privacy risks incurred by the integration of AI and IoT with the following research tasks: 1) building a joint research testbed for integrating AI and IoT; 2) investigating data poisoning attacks and defenses on AIoT devices; 3) enhancing model privacy and robustness for federated learning in AIoT; and 4) empowering AIoT through secure computation offloading. Through these tasks, this project will result in novel research discoveries in the area of AIoT, including understanding to latest AIoT attacks and countermeasures. Students at York, majority of whom are from minority groups, will be recruited to engage in the research. New courses and modules will be developed in this project. The project outcomes will be disseminated to the community through website, workshops, journals and international conferences. 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 2024 · 2024-11

Oceans cover over 70% of the Earth's surface and their ceaseless movement from tides, waves, and currents creates a potentially important energy source that could be an important component of the energy transition for coastal and island communities. Similarly, the constant cycling of onshore and offshore winds over the course of a 24-hour period creates an additional marine related source of green, renewable energy. The Industry-University Cooperative Research Center (IUCRC), the Center for Growing Ocean Energy Technologies and the Blue Economy (GO Blue) engages in faculty-driven/industry-relevant, basic, use-inspired research focused on creating new knowledge and innovations of value to industries and start-ups in the marine energy ecosystem. Created by three universities: the University of Michigan, Ann Arbor; the Stevens Institute of Technology; and Texas A&M University at Corpus Christie, this national Center has the potential to address critical problems and issues that are holding the economy back from realizing economically viable electricity coming from marine and coastal marine-related energy sources that generate electricity to feed the national power grid. Broader impacts of the Center include the creation of new knowledge and solving of technical problems and socio-economic issues associated with marine electrical energy generation, close collaboration between university faculty and students and industry, training students in innovation; entrepreneurship; and workforce and workplace safety, and developing new educational programs to build the marine energy workforce of the future. Other impacts include public outreach and community engagement around marine energy issues. The Go Blue Industry-University Cooperative Research Center will harness the expertise of over 30 faculty and researchers from three universities and engage their students and postdocs in basic but industry-need-inspired research. Engineering research in this center will be inherently multidisciplinary spanning the fields of electrical, mechanical, civil, ocean, materials, and environmental engineering.This multi-university collaboration provides expanded access to world-class schools of naval architecture and engineering, state-of-the-art ocean technology test facilities, and computational facilities to Center faculty and students, regardless of home institution, as well as to members of the Center Advisory Board through faculty-initiated research projects of high priority to the marine energy economic sector. The geographical distribution of the three partner universities: The Great Lakes (Michigan), ocean (Stevens Institute), Gulf of Mexico (Texas A&M Corpus Christi), creates a national ensemble that allows the Center to tackle and experiment with new ideas, technology, and energy implementation scenarios in vastly different marine/coastal/large freshwater lake environments and settings. Center research ideas come from faculty who listen to the needs of the marine energy economic sector, as represented by dues-paying members of the Center Advisory Board. Faculty research is funded by pooled Advisory Board membership fees for projects of high priority to the sector, as indicated by the collective members of the Center Advisory Board. Center research thrusts are marine energy technology for renewable energy production, powering the blue economy which includes power generation and marine transport, and marine energy societal acceptance; economic viability; and environmental sustainability. 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 2024 · 2024-10

Unit testing, a fundamental phase of software testing, faces challenges due to interdependencies among the function units in a system. It is impractical to test a system as completely separate units without considering their dependencies. Software engineers have devised a mechanism called mocking, which replaces the test dependencies of the function under test by creating mock objects. Various dedicated mocking frameworks provide well-constructed application programming interfaces (APIs) to support the mocking mechanism in practice. However, practitioners often find adopting these frameworks challenging. This award targets the following challenges: (1) there is a knowledge gap regarding what to mock and what not to mock in unit testing; (2) despite mocking being used in practice for decades, there is little understanding of good practices to follow and bad practices to avoid in mocking; and (3) mocking and its maintenance require a high level of expertise and manual effort, which raises the bar for using mocking frameworks. This award aims to contribute valuable empirical knowledge and develop a framework with automated support for achieving "just enough" and "maintainable" mocking in unit testing. The proposed research will facilitate creating more independent, efficient, easier to debug, and more maintainable unit test cases. The research team also plans to develop an advanced course about mocking for undergraduate seniors and graduate students in Software Engineering, Computer Science, or related fields. Additionally, the team plans to prepare publicly available online tutorials and training sessions to benefit practitioners interested in building skills in mocking. The research team will focus on three research directions in this project. First, conducting an in-depth empirical study of mocking practices. The objective is to provide extensive empirical knowledge about what to mock and what not to mock, including good "design patterns" to follow and bad "anti-patterns" to avoid in mocking. Such empirical knowledge benefits practitioners by offering guidance and references for effectively using mocking framework APIs. Second, developing a framework to enable "just enough" and "maintainable" mocking. This framework consists of two key components: (1) automated detection of dependencies to mock to achieve "just enough" mocking; and (2) automated detection and refactoring of mocking "anti-patterns" to ensure good design and maintainability of mock objects in test cases. Finally, developing curriculum and training materials about mocking. 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 2024 · 2024-10

In principle the highest assurance of trustworthiness for software systems is achieved by mathematical proofs of correctness with respect to precise specifications of requirements. Specifications describe the structure and intended behavior of software based on mathematical models of hardware, software, and the operating environment. The environment may include active adversaries trying steal or tamper with sensitive information. Creating specifications and proofs is difficult and is seldom done in current practice, but some companies have begun to adopt so-called auto-active verification tools. An auto-active verification tool enables the software engineer to produce specifications and proofs by annotating code with special comments called assertions. Assertions are automatically checked, and the tool is integrated with commonplace development tools and workflows. This approach works well for requirements that can be expressed in terms of individual executions of a program. However, confidentiality and privacy requirements must be specified as relationships between multiple executions. Relationships are also needed to verify equivalence between different versions of a program. Such relational requirements are called hyperproperties. In this project, researchers are developing theory and algorithms to extend the reach of auto-active verification to include hyperproperties. The research has the potential to transform computing practice by supporting accountability of system designers and builders, and trustworthiness of software supply chains, through evidence that includes mathematically precise specifications and machine-checked proofs of security and privacy. Usable equivalence verification can lessen the risk that code improvements and bug fixes introduce new vulnerabilities. The project plan is organized in three thrusts. One thrust is advancing the science of security by developing theory in the form of deductive logics together with novel algebras to support automated reasoning by alignment of intermediate steps in relationships between executions. Deductive rules are connected with annotation-oriented verification conditions and also with code-independent confidentiality and integrity properties specified with respect to a model of adversary capabilities. The second thrust is developing algorithms for automated checking of annotations and search for alignments that facilitate annotation. Existing prototype tools are being extended with these algorithms. Trustworthiness of the tools is assured by the theory developed in the first thrust, which is being machine-checked using an interactive proof assistant. In the third thrust, experiments and case studies using the prototypes serve to evaluate usability and effectiveness of verification for requirements including secure information flow, program refinement, and program equivalence. Research products including training materials and benchmarks are being made freely available. 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 2024 · 2024-10

Federated Learning (FL) has emerged as a popular distributed machine learning paradigm in a wide range of sectors (e.g., healthcare, fintech, and autonomous driving) because of its potential of protecting people’s privacy - it does not require gathering all the data in one place for operation. Meanwhile, driven by the increasing ubiquity of mobile devices, FL applications are shifting from wall-plug powered artificial intelligence (AI) devices to battery-powered mobile AI systems (e.g., smartphones, tablets, wearables). Existing research largely ignore the role battery energy awareness plays in efficient FL training over mobile AI systems. This project addresses this challenge and innovates on developing an energy-efficient FL framework for mobile AI systems, making these systems more suitable for execution on everyday mobile devices without draining their batteries quickly. The project's broader significance and importance are its potential to advance mobile computing and AI technologies, ensuring both are energy-efficient and privacy-preserving. Furthermore, this project shares its research artifacts and results with the community and includes educational activities targeting under-represented groups in computing. This project investigates the efficiency, quality, and robustness of FL systems from an energy perspective, aiming to develop a comprehensive energy-efficient FL framework for mobile AI systems. The research is structured around three synergistic objectives. First, the project develops a universal energy estimation methodology applicable across a variety of devices engaged in FL training, incorporating Deep Neural Network (DNN) models with diverse architectures. Next, utilizing insights into energy consumption, the project explores strategies to enhance the energy efficiency of FL, particularly in high-speed communication scenarios such as autonomous driving and augmented/virtual reality. Additionally, the project integrates learning performance metrics, such as accuracy and latency, with energy parameters—including energy consumption and battery life—in the FL participant selection process. This integration aims to create a balanced and optimized learning environment. To support these goals, the project establishes a mobile AI testbed and energy measurement setup, equipped with real-world FL benchmarks and workloads. 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 2024 · 2024-10

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Next-generation serverless cloud computing provides developers with simplified access to server management and administration, including event-driven execution, fine-grained resource provisioning, auto-scaling, and pay-as-you-go billing. The machine learning community is taking advantage of these benefits of serverless cloud computing to ease the development and deployment of deep learning (DL) applications. However, existing serverless computing platforms lack efficient support for GPUs, impeding DL practitioners from utilizing serverless computing for large-scale applications. This project will develop an efficient serverless computing platform with a hybrid CPU/GPU architecture to accelerate DL application development and deployment. The goal is to advance cutting-edge methodologies in both deep learning and serverless computing, which will result in a significant leap forward to benefit DL practitioners, DL users, and providers of cloud computing infrastructures, contributing to science advancement for society. The research findings will also enhance undergraduate and graduate education with exciting examples and demonstrations of real-world systems at the intersection of distributed computing, cloud computing, and deep learning. The project will develop a novel serverless computing platform with a hybrid CPU/GPU architecture that will provide DL applications with native GPU performance. Two core components constitute the hybrid serverless computing architecture, a shim virtualized GPU (vGPU) layer and a refactored container subsystem. The shim vGPU layer enables high-performance GPU sharing for concurrent serverless functions with low latency and high scalability. This layer provides fine-grained GPU resource provisioning and performance isolation by intercepting GPU calls from serverless functions using API remoting techniques. The vGPU layer optimizes GPU performance in serverless computing via GPU context caching and locality-aware scheduling to mitigate cold-starts and unnecessary data movement. The container subsystem accelerates the entire DL lifecycle by exploiting DL model structures and pipelined model loading to parallelize CPU-to-GPU memory copy and model execution. The subsystem exploits model partitioning techniques to accelerate the hybrid CPU/GPU architecture by dynamically distributing the DL model partitions to CPU and GPU. The scientific knowledge and tools designed and implemented from this research project will provide and enable innovations for next-generation cloud computing and deep learning. 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 2024 · 2024-10

Supercomputers, or high-performance computing (HPC) clusters, are instrumental in propelling scientific and engineering research by offering vast computational resources. These systems are increasingly crucial as artificial intelligence (AI) techniques become pervasive across various fields, including climate modeling, drug discovery, and physics simulations, significantly expanding the need for computational power and data management. However, the existing HPC infrastructures face challenges with extended job wait times and suboptimal resource use, primarily due to the escalating complexity of computations and the burgeoning demands for resources. Unlike traditional HPC tasks, AI algorithms and models exhibit distinct resource requirements, often resulting in either excess or insufficient resource allocation for AI tasks. This project aims to bridge the gap between HPC resource provisioning and AI application demands through an in-depth analysis of resource allocation and utilization within HPC environments running AI workloads. The goal is to identify strategies for minimizing resource waste and reducing the length of job queues by efficiently reallocating idle resources to accommodate large-scale AI tasks. By creating and disseminating datasets, models, algorithms, and system source code, this initiative will contribute valuable tools and insights to the research community. The findings will be broadly shared through research papers, technical reports, book chapters, course materials, and tutorials, enhancing the knowledge base in both HPC and AI fields and supporting the broader objectives of promoting scientific progress, improving national health, prosperity, and welfare, and contributing to national defense. This project centers on advancing the efficiency and productivity of HPC systems by innovatively leveraging idle resources to expedite AI job processing and diminish waiting periods. The research is structured around three interconnected themes, each addressing critical aspects of resource utilization and AI performance enhancement within HPC environments. The initial theme undertakes a comprehensive analysis of idle resources in HPC systems, aiming to identify patterns and opportunities for resource optimization. Building on the insights gained, the second theme explores methodologies for the safe and timely harvesting of idle resources across various categories, ensuring that these resources can be reallocated without compromising system stability or performance. The third theme is dedicated to developing strategies that utilize these harvested resources to boost AI application outcomes significantly and, by extension, enhance the overall productivity of HPC operations. The project will implement a tangible HPC testbed equipped with real-world benchmarks and workloads alongside these thematic investigations. This testbed will serve as a platform for empirically validating developed algorithms and systems, facilitating a rigorous assessment of their effectiveness in improving HPC resource allocation and utilization. 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 2024 · 2024-10

Access to comprehensive knowledge about diseases, conditions, and medications is not only empowering but also essential for the public to understand complex medical information, leading to better personal well-being. However, current medical information sources, such as Google Knowledge Graph, often have limited scope, primarily covering a small subset of well-known diseases. Public data sources like academic papers on uncommon diseases are not well-structured or easily understood by the general public. This project aims to create a flexible and open-resource medical knowledge base called FORMED, providing multi-faceted information on a wide range of diseases and conditions for public access. This knowledge base will include well-structured sections on symptoms, causes, and treatments, enabling efficient disease classification and indexing. By integrating medical knowledge with individual health records, the project will also evaluate its effectiveness in predicting individual health risks for uncommon diseases. Additionally, this project will involve educational initiatives such as developing new courses on large language models; conducting interdisciplinary research activities to train graduate, undergraduate, and high-school students in data science and bioinformatics; and increasing participation of women and minority groups in academic research. All core outcomes of this project, including software, datasets, and publications, will be made available to the general public. The goal of this project is twofold: (1) to create a public-oriented medical knowledge base called FORMED, covering a wide range of diseases, conditions, and medications in the current disease classification system with descriptive attributes including symptoms, causes, and treatments; and (2) to develop a temporal health outcome prediction and generation framework to evaluate the generated knowledge base with individual health records. This project creates a set of technologies for semi-structured text generation, knowledge graph construction, and mixed-structure temporal data prediction as well as generation. Specifically, the research activities include: (1) Developing hyperbolic embedding-enhanced domain-specific large language models for building FORMED; (2) Constructing a knowledge graph from FORMED to represent the logical concepts of disease characteristics and causes; (3) Designing novel learning and prompting strategies to augment the reasoning capability of large language models with knowledge graphs; and (4) Building a robust testing platform to evaluate the effectiveness of the generated knowledge graph for forecasting individual health risks. The establishment of comprehensive medical knowledge bases will significantly enhance public understanding of uncommon diseases and improve the inference capabilities of generative models, enabling searching-based services. Research outcomes of this project will be disseminated in peer-reviewed publications, tutorials, seminars, and workshops. 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 2024 · 2024-10

Emerging polymer technologies include biogels, purification membranes, recyclable plastics, and advanced composites. However, progress in these areas is hindered by insufficient tooling for preparing and analyzing polymer simulations. This work leverages expertise in high-throughput computing, polymer physics, reaction dynamics, and scientific software development to enable efficient, reproducible modeling across multiple simulation engines and hardware architectures. The scientific and software needs of universities, national labs (NREL, LLNL, INL, NIST, AFRL), industry (Boeing, Bristol Myers Squibb) and international consortia (CECAM, FairMAT) are incorporated to maximize impact. The Multiscale Polymer Toolkit (MuPT) enables reproducible and extensible computational research on reacting polymer materials from Angström to micron length scales. MuPT is an expanding suite of Python software libraries and community recipes, built on top of an ecosystem of previously funded open-source tools. The effort pairs Open Molecular Software Foundation software developers with domain experts to develop software and tutorials in collaboration with application scientists in the community. Findability and accessibility are accomplished through conda-forge deployment and public workshops. MuPT deliverables include: (a) A multiscale, internal software representation for polymers that enables data conversion between major simulation engines at the same resolution scale, and tools for conversion between coarse-grained and higher resolution representations; (b) An interface for this representation that allows researchers to plug in existing software tools for polymer parameterization, building, and crosslinking; (c) A workflow interface that allows linking of existing software tools and enables users to programmatically generate simulation inputs by specifying the simulation engine, chemistries, reaction models, and molecular representations; (d) A searchable repository of community-vetted polymer simulation workflows, initially seeded and maintained by the principal investigators of the grant; (e) Documentation for best practice in polymer modeling with examples using MuPT libraries; (f) Improved materials and recommendations for training research software engineers. This award by the NSF Office of Advanced Cyberinfrastructure is jointly supported by the Division of Materials Research. 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 2024 · 2024-09

Civil and construction engineering students may encounter various construction components, including structural elements, materials, equipment, and operations, in their everyday lives, such as when walking in an urban environment. These unstructured observations can offer great learning opportunities; however, without expert support, it is unlikely that students will effectively learn from such observations on their own. If educators were physically available during students' everyday activities, they could direct students’ attention to the main construction components and explain their observations in real-time. However, since this is not feasible in the real world, this project aims to design, develop, and test a transformative learning system that uses Artificial Intelligence (AI) as an on-demand educator. The envisioned AI-enhanced learning system relies on a digital platform in the form of a mobile application. When students face a construction project, they can look at the project through their smartphones using the mobile app, which will help students learn from their observations by 1) directing their attention to the main construction components they encounter in their everyday life or formal site visits, 2) explaining the observations, 3) linking the observations to students’ formal engineering education materials available on web-based learning management systems, and 4) generating automated reports about students’ observations and performance for instructors to help them adjust the course activities accordingly. To promote equity and accessibility in education, the mobile app will be designed to operate on the most basic and affordable smartphones and will use color palettes compatible with the needs of users with color vision deficiency (CVD), along with subtitles and audio narrations. The envisioned AI-enhanced learning system will be designed based on the Activity Learning Theory, which asserts that the human mind is an integral part of environmental interactions and positions activity—whether sensory, mental, or physical—as a precursor to learning. The AI-enhanced platform will be designed based on human-centered principles and will operate using a novel hybrid image-audio processing system that can efficiently and effectively recognize and classify various construction components. In addition to integrating imagery and audio data through this novel hybrid approach, the project will introduce two major technological innovations in audio processing and sound recognition. First, the hybrid use of collected audio and imagery data will improve the overall performance of the system by capturing a more comprehensive range of construction components and operations. Second, by using innovative audio processing and signal source separation algorithms, the need for multiple microphones will be eliminated, enabling the entire system to be encapsulated in a single device (i.e., a student’s smartphone) with the ability to sense and analyze audio signals from distances of up to 100 feet. Throughout this project, the proposed AI-enhanced teaching and learning approach will be implemented in multiple undergraduate construction engineering courses to empirically evaluate its effectiveness on students’ learning processes and outcomes, as well as the perceptions of both students and educators regarding this innovation as a formal pedagogical method. Although the AI-enhanced learning platform will be developed in the context of construction engineering, the proposed learning method and the intellectual merit of this project can be transferred to other disciplines. This project will also assess the broader applicability of the proposed innovation. This project is funded by the Research on Innovative Technologies for Enhanced Learning (RITEL) program that supports early-stage exploratory research in emerging technologies for teaching and learning. 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 2024 · 2024-09

Modern integrated circuits (ICs) integrate billions of transistors into a single chip. Even key subsystems such as clock distribution networks, power delivery networks (PDNs), embedded memory arrays, and analog and mixed-signal systems can become incredibly complex, consisting of hundreds of millions of circuit components. As a consequence, it has become exceedingly difficult to model, simulate, optimize, and verify future ICs at a large scale using existing design methodologies. Recent research in spectral graph sparsification has made it possible to construct much sparser subgraphs while preserving important graph spectral properties such as the first few eigenvalues and eigenvectors of the graph Laplacian. These findings have already led to the development of highly efficient algorithms for solving large sparse matrices, partial differential equations (PDEs), and simulating large-scale circuit systems. However, existing spectral sparsification methods do not efficiently allow for updating the sparsified graph when only incremental changes are made to the original network. The algorithms and methodologies to be developed in this project will be transferred to technology companies, such as semiconductor and electronic design automation (EDA) companies, as well as social and network companies for potential industrial applications. Furthermore, the source code of the developed EDA algorithms will be made available to other researchers through collaborations. The project also includes compelling education, outreach, and course development plan. The primary research investigator plans to recruit K-12 students from local schools and get them involved in the latest research projects. This research project aims to investigate scalable algorithms for incremental spectral sparsification of large graph Laplacians and integrated circuit networks. The framework will allow for efficiently handling large graphs that experience streaming edge insertions/deletions in sublinear time, and significantly accelerating SPICE-accurate modeling and simulations of large circuit designs that undergo frequent updates. This research will also investigate efficient approaches for incremental modeling and simulation of integrated circuits leveraging incremental spectral sparsification and involve close collaborations with leading industrial experts in circuit simulation to evaluate the proposed research. The success of the proposed research will potentially contribute to the advancements in spectral graph theory and EDA, leading to the development of highly scalable incremental algorithms for solving partial differential equations, sparse matrix problems, spectral graph partitioning, and design automation of large-scale ICs. This project is co-funded by the Software and Hardware Foundations (SHF) and Discovery Research PreK-12 (DRK12) programs. DRK12 is an applied research program that supports STEM education PreK-12. 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 2024 · 2024-09

Solute dispersion in unsaturated porous media has critical implications for various natural and engineered systems, such as nutrient and contaminant transport in the vadose zone, the unsaturated region between the ground surface and the water table. However, relatively few studies have been conducted on solute dispersion in unsaturated porous media compared to in saturated porous media. This is partially due to the challenges in studying multiphase flows (i.e., co-flow of immiscible fluids) in porous media. The permeability of a porous medium describes the medium’s capability to allow a fluid to flow through it. The permeability of natural geological formations is usually complex, which dictates the solute transport mechanisms in unsaturated flows. Experimental data concerning the role of complex permeability on solute dispersion in unsaturated flows are extremely rare due to the challenges encountered by conventional experimental methods in constructing a well-controlled permeability field in the laboratory. The objective of the research is to use pore-scale numerical simulation, microfluidics, and three-dimensional (3D) printing technologies to overcome the challenges encountered by conventional experimental methods in order to advance the understanding of the impact of complex permeability on solute dispersion in unsaturated porous media. The educational objective is to increase public scientific literacy, engage women and minority students in STEM, enrich undergraduate and graduate curricula, and develop partnerships with the industry and small businesses. A fundamental knowledge gap exists in how the dynamic interaction between fluid co-flow and complex permeability regulates solute dispersion in unsaturated flows. The project aims to close the knowledge gap using pore-scale numerical simulation and state-of-the-art 3D printing. The research tasks are: 1) develop a pore-scale numerical model to directly simulate two-phase flow and solute dispersion in porous media to unravel the impact of water saturation, Peclet number, and capillary number on the dispersivity and develop an empirical model for dispersivity predictions, 2) develop an advanced observation method for in situ solute concentration measurements and conduct two-fluid co-injection and solute dispersion experiments in a microfluidic device to validate the pore-scale simulation results, and 3) use 3D printing to construct a known and well-controlled heterogeneous permeability field at the continuum (meter) scale to study its role on solute dispersion in unsaturated flows. 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 2024 · 2024-08

Real-world applications, such as software modeling, digital circuit design, manufacturing control, and status modeling of smart devices and smart systems, often require efficient techniques to model their behaviors and changes over time. Based on their specific requirements, different algorithms (including machine learning) are needed, such as reachability computation, pathfinding, and state prediction. For example, the graph neural network (GNN) algorithm can help to learn the compact vector representations of the states and transitions to capture the complex patterns and dependencies. However, existing computation architectures for such techniques are not very efficient for two major reasons: (i) the algorithms are not computationally efficient, and (ii) the data size is very large. This research pioneers the development of an accelerated computation architecture for system modeling techniques and applying them to critical smart environment applications. This project will address the growing national need for professionals in accelerated computation architecture, algorithms, and machine learning. The research will produce an accelerated computation architecture that serves as a foundational tool for fellow science and engineering practitioners in academia and industry. Educational initiatives integrate the research findings into graduate and undergraduate curriculum development. Additionally, outreach and educational activities are conducted to promote participation from K-12 and undergraduate students from populations underrepresented in computing. The overarching goal of this project is to design an accelerated computation architecture for state modeling techniques and to apply them to important smart environment applications. Towards that, this project includes three synergistic research thrusts. Specifically, Thrust 1 designs efficient computation techniques to accelerate the reachability computation in a state transition representation, which can be used to detect if any undesired (e.g., unsafe) state is reachable. Thrust 2 accelerates the computation of graph machine learning algorithms by adaptively reducing the overhead of instant updates and maintaining high-quality communities. Thrust 3 applies the techniques in Thrusts 1 and 2 to an important application domain of smart environments. 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 2024 · 2024-07

The national spectrum strategy emphasizes spectrum infrastructure and workforce development in the full range of operational, technical, and policy roles to establish U.S. leadership in future-generation (FutureG) wireless techniques. However, achieving the strategic goal of spectrum workforce development involves non-trivial challenges, including limited capacity and availability of advanced wireless cyberinfrastructure (CI) and specialized tools, skills, and knowledge sets to develop, manage, and utilize wireless CI. This project responds to the national call for spectrum workforce development and trains the FutureG workforce by extending their research abilities through a novel “immersed” approach to promote project-based hands-on learning. An open radio access network (O-RAN) wireless testbed will be utilized to allow trainees to practice the operation and programming of FutureG wireless instruments and develop wireless applications. Cloud-based access to the O-RAN testbed and a suite of template projects will be offered to address the technical barriers and complexities of wireless CI access. New course modules, vertical-integration projects, and summer courses will be offered to both student trainees and existing students at PIs’ institutions. The training materials will be disseminated through public platforms including the ACCESS Knowledge Base to train a broader and diverse group of wireless professionals. Specifically, this pilot project includes three tasks: Task 1 is to extend the abilities of wireless professionals with a publicly and remotely accessible wireless CI based on the O-RAN architecture. The CI integrates advanced RF and computing instruments including NI USRP X410 and 2974 supporting sub-6G Hz to mmWave bands, TMYTEK mmWave BBox at 28GHz and 39GHz, phrased-array beamformer and reconfigurable intelligent surface (RIS), and a GPU server with 8x NVIDIA RTX A5000. Task 2 aims at training wireless professionals with the development of AI/ML tools and services to allow automatic wireless data collection and intelligent analytics. Based on this unique CI, Task 3 develops a suite of hands-on projects to train and educate wireless professionals under different scenarios ranging from basic wireless instrument operation to advanced wireless research. 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 2024 · 2024-07

This project advances our knowledge of how to develop and implement transactive energy services in rural communities. A transactive energy service is an energy-sharing, real-time pricing market that exchanges electricity among retail customers and power producers that could include large generation companies and households with rooftop or backyard solar. Transactive energy services have long been recognized as potentially useful for decarbonizing the electric grid, enhancing community resilience to supply shortages, extreme weather, and natural disasters, and reducing electricity rates for everyday consumers. This project aims to assess the social and technical factors influencing the potential for implementing transactive energy services. The project is investigating the potential for a transactive energy service in rural New Hampshire with a collaboration of community power partners. Through surveys, focus groups, and interviews, the research team is studying the technical and social psychological factors that will influence rural New Hampshire residents' willingness to enroll in the transactive energy service. It is examining the technical and social features and factors that can produce a market that is socially beneficial, technically reliable, and environmentally sustainable. And it is examining human-machine interactions to produce an artificially-intelligent in-home automation system that facilitates rural New Hampshire residents' participation in the transactive energy service. The project is also assessing if the knowledge of transactive energy service design and implementation generated in New Hampshire could be scaled to, transferred to, and sustained by other community partnerships or coalitions in the United States. 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 2024 · 2024-07

Data in various domains typically exhibit multiple levels of abstraction, ranging from representations based on language (i.e., semantic representations) to low-level sample-specific details. Examples include in facial images, range from overall facial structure to specific features like eyes and mouth, down to textures like skin and hair, and the layered organization in language data, from topics to paragraphs, sentences, and words. Learning and understanding these data to effectively generate novel content presents a challenge for modern artificial intelligence (AI). Generative AI, which has garnered increased attention in recent years, provides principled methods to begin to address this challenge. Unfortunately, existing models often overlook the structural information within data. They also lack controllability because much of what happens is not transparent. Finally, many models are domain-specific, limiting their capacity in new areas and hindering their use in safety-critical and cross-domain applications. This project aims to go beyond existing generative frameworks by developing new models that exceed their current capacities while maintaining the ease of controllability. The project will also support curriculum development for both graduate and undergraduate programs in artificial intelligence. Furthermore, the principal investigator will continue to mentor undergraduate and graduate students and will be actively involved in pre-college programs for K-12 STEM education. The technical objective of this project is to design new structured generative modeling and learning frameworks to advance existing artificial intelligence for more informative, controllable, and cross-domain representations. Specifically, this project will study and establish core methodological foundations for hierarchical generative modeling in three key thrusts. The first thrust develops context-aware generative models that incorporate the modeling of structural information and contextual dependencies for expressive hierarchical representation. The second thrust develops new structured and semantic-inducing schemes for controllable models. The third thrust develops multimodal generative models for effective cross-domain representations. The comprehensive investigation of the project will lead to the design of more powerful and reliable generative artificial intelligence systems that are easily understandable and effectively manageable by users. 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 2024 · 2024-06

This award supports participation in the conference “Groups, Logic, and Computation: Interactions between group theory, model theory, and computer science" that will be held at the Stevens Institute of Technology (New Jersey), June 12-14, 2024. The event will bring researchers from various branches of group theory, model theory, and computer science together to work on some of the many open questions in the field that are now being studied from fresh and promising perspectives; it will further strengthen the connections the field has to other branches of mathematics. This exchange of ideas among experts, students, and postdocs aims to disseminate current knowledge and identify promising directions for further progress. The conference will be devoted to developments in group theory focusing on groups and group actions as well as other areas of mathematics in which groups or group actions are used as a main tool. The program covers many branches of modern group theory with preference given to geometric, asymptotic, and combinatorial group theory, dynamics of group actions, probabilistic and analytic methods, first-order rigidity, first-order classification, and Diophantine problems in groups and rings. More information can be found at https://web.stevens.edu/algebraic/Stevens2024/ 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 2024 · 2024-06

Machine learning services are quickly rising in popularity, allowing companies and individuals to reap the benefits of efficient and accurate deep learning models. As a result, these systems are handling an increasing amount of data, including potentially sensitive user data. Concurrently, architectures such as adaptive neural networks show promise in addressing the high memory and workload requirements of deep neural networks. However, these adaptive neural networks are susceptible to a class of vulnerabilities called timing side channels through which timing observations are leveraged to learn sensitive information about private user input. The project's novelties are providing the first investigation into the potential of adaptive neural networks to leak private user information through timing channels and developing strategies to effectively mitigate this leakage while still reaping the benefits of more efficient adaptive neural networks. The project's broader significance and importance are enabling the development of adaptive neural networks able to balance privacy requirements with performance demands, allowing them to be confidently deployed on devices such as mobile phones or small smart devices. This project provides the first holistic and systematic approach to understanding timing side channels in adaptive neural networks. The contributions of the project include: (1) defining a threat model under which timing side channels can be exploited by an attacker to gain sensitive information, (2) a machine-learning-based pipeline to effectively utilize the timing channels under this attack model, (3) an automated fuzzing-based approach to guide development of models towards those without potential timing channels, and (4) an online monitoring framework to dynamically check the balance of privacy, accuracy, and performance in deployed systems. The approaches in this project are generalizable across a variety of types of user data and adaptive neural network architectures, and the work will lead to advancements in automated testing of machine learning frameworks for violations of user privacy. 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.