Temple University

NSF Awards · FY 2024 · 2024-09

This project aims to enhance how computers understand and recognize shapes in images and 3D data. While current advanced computer systems are powerful, they struggle to perceive objects as humans do—as whole shapes made up of interconnected parts. This project seeks to teach computers to focus on the critical parts of an image that constitute an object, disregarding the background and other distractions. The project is also developing methods for computers to comprehend the relationships between different parts of an object, akin to recognizing a chair by identifying its seat, legs, and back in a specific arrangement. This will be achieved within recent deep learning frameworks without explicitly referencing such parts. The improved visual perception capabilities could help computers better to perform tasks like object recognition, 3D position estimation, and understanding shapes from various viewpoints. The potential applications of this research are vast, including advancements in medical imaging, dental radiology, augmented reality, assisted driving technologies, and robotics. Additionally, the project will involve students from high school to doctoral programs in research, with a focus on including underrepresented groups. The project creates a framework for deep learning models to learn configural (or holistic) shape representation and arrangements of object parts, demonstrating their applicability to various computer vision tasks. It will augment the learning framework of vision transformers to learn how to restrict the attention of image patches. This is achieved by adding a new branch that computes an auxiliary loss called object-focused attention, which limits the attention to patches belonging to the same object. This allows transformers to gain a better understanding of configural object shapes by largely ignoring the background and other objects. Additionally, the project will develop a novel graph-transformer-based shape configuration framework named ShapeGT for generic shape understanding tasks. ShapeGT will include several new techniques and applications, including (1) novel between-edge attention and edge-to-node attention modules, (2) joint graph learning and matching algorithms, (3) view encoding techniques for multi-view analysis, and (4) cross-modal fusion mechanisms for capturing 3D-to-2D (shape-to-image) interactions. Due to the wide range of applications of shape analysis, the project is expected to have broader impacts on fields beyond computer vision, such as dental maxillofacial radiology, augmented reality, molecular biology, assistive driving, medical image analysis, and robotics. 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

About 15-20% of the adult population identifies as neurodivergent. These individuals offer immense unrealized potential as employees; however, they are a vulnerable community subject to extreme social and systemic inequities in jobs and higher education. Neurodivergent adults experience chronic unemployment and underemployment. When employed, they are underrepresented in management and leadership roles within organizations. Solving this complex problem goes far beyond the reach of any single discipline, but requires theories, methodologies, and approaches that encompass policy, organizational, group, individual and technological insights as well as meaningful involvement of neurodivergent individuals. The objective of this planning proposal is to assemble a team of researchers, employers, educators, advocates, and neurodiverse individuals to study, develop and disseminate organizational and technological evidence-based practices to better support the advancement of neurodivergent individuals in meaningful, productive work and increase worker productivity, job satisfaction, and career advancement. This project brings together an interdisciplinary team of researchers with expertise in artificial intelligence, behavioral science, data science, game design, organizational psychology, physical therapy, rehabilitation science and special education, to collaborate with advocates, educators, employers, and neurodivergent individuals to transform the current state of employment for adults who identify as neurodivergent. Building on previous NSF funded research, the work described in this planning proposal will create a muti-university, multidisciplinary Center for Neurodiversity Development and Advancement that includes both researchers and key stakeholders collaboratively designing research questions and developing solutions. Integrating scientific knowledge from educational, organizational, technological, and psychological research, each participating university capitalizes on its unique strengths and builds a collaborative team with neurodivergent individuals and advocates included as partners. Products of the center will include research to solidify factors underlying lack of employment opportunities, development of supports to enhance access to higher education and job training in collaboration with the neurodivergent community, development of strategies to facilitate meaningful employment and career advancement, and education to organizations and other key stakeholders within the broader community to promote employment equity. 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

Decentralized Federated Learning (DFL) has emerged as a new learning paradigm in artificial intelligence, enabling the training of data-hungry learning models on local devices without sharing raw data. This paradigm is immune to the single-point failure of a central server and the privacy perils caused by a dishonest server. However, the understanding of DFL is still in its infancy. It is unclear how the decentralized and periodic communication strategy affects the convergence performance of DFL algorithms, especially when tackling emerging machine learning models where the corresponding optimization problem has complicated structures, such as bilevel optimization. Furthermore, peer-to-peer communication in DFL introduces unique security risks, stemming from a combination of malicious users and device-to-device communication patterns. This project aims to design and develop a secure and efficient DFL system, addressing these communication, computation, and security issues. This project will benefit a variety of high-impact applications where machine learning models are trained in a DFL setting without sharing raw data. This project aims to develop computational theories, models, and prototype systems, forming the foundations for trustworthy DFL, considering both high-performance accuracy and security with privacy preserving. The first focus of research is to develop the structural communication topology and pattern, underpinned by mathematical graph theories and empirical computer network techniques, to favor efficient and robust communication. The second focus is to investigate the bilevel optimization problem for emerging machine learning models in DFL, where efficient stochastic bilevel optimization algorithms will be developed, and their theoretical convergence foundations in DFL will be established. To providing security guarantees, unique security threats to DFL will be thoroughly investigated and principled defense strategies will be developed accordingly. Beyond these foundational aspects, this project will apply the developed techniques to practical data mining applications in Internet-of-Things networks and Smart Transportation, addressing the unique challenges therein and providing practical solutions to benefit real-world applications. Moreover, the team will integrate the proposed research work into several courses and provide abundant research activities for both undergraduate and graduate students with diverse backgrounds. 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

The National Science Foundation (NSF) Graduate Research Fellowship Program (GRFP) is a highly competitive, federal fellowship program. GRFP helps ensure the vitality and diversity of the scientific and engineering workforce of the United States. The program recognizes and supports outstanding graduate students who are pursuing research-based master's and doctoral degrees in science, technology, engineering, and mathematics (STEM) and in STEM education. The GRFP provides three years of financial support for the graduate education of individuals who have demonstrated their potential for significant research achievements in STEM and STEM education. This award supports the NSF Graduate Fellows pursuing graduate education at this GRFP institution. 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

A 3-manifold is a space where an object can move around in three distinct perpendicular directions. The universe is a three-manifold whose global structure is not yet understand. Thanks to transformative work around the turn of the century, it is known that the geometry of a manifold (measurements of angles, distances, and curvature) is closely tied to its large-scale structure. What is missing at this point is a quantitative understanding of how geometry and large-scale topology determine one another. This project seeks quantitative information of this nature. The project includes problems pursued in collaboration with current and recent graduate students mentored by the PI. The project also supports the PI’s leadership efforts in building stronger mentoring in his department, nurturing the mathematical community in the Philadelphia area, and training junior mathematicians through a national graduate conference. Mathematically, this project seeks to make progress on several important open questions about hyperbolic 3-manifolds and their fundamental groups, with emphasis on effective computation and large-scale structure. One question involves quantitative control on the change in geometry under Dehn surgery, including applications to the cosmetic surgery conjecture that are coded into software for the mathematical community. The second question involves identifying the Margulis constant and understanding the structure of Kleinian groups generated by two short elements. The third question involves relationships between the rank and genus of 3-manifolds. The fourth question involves a coarse understanding of the fixed-point properties of pseudo-Anosov maps on surfaces, with applications to invariants of knots and 3-manifolds. The fifth question involves a quantitative understanding of special covers of 3-manifolds. 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

Artificial Intelligence (AI) has reached groundbreaking milestones in recent years. Its usage has spanned critical application domains, such as computer vision, audio perception, and natural language processing. However, these breakthroughs come with substantial security challenges. The machine learning (ML) models serving as the computational cores of AI systems are inherently vulnerable to attacks. By exploiting vulnerabilities in AI systems, adversaries can make the models produce incorrect predictions, leading to serious consequences such as misinterpreting traffic signs for autonomous vehicles or generating incorrect responses in speech recognition systems. Current AI-related educational efforts are limited on teaching the security perspective of ML. To bridge this gap, this project aims to develop comprehensive educational modules to prepare students and future engineers to address these ML security vulnerabilities and achieve trustworthy AI. By creating a practice-in-the-loop learning experience, students can obtain hands-on experiences with the security vulnerabilities of ML models and corresponding solutions. This project will develop a comprehensive educational program that focuses on three key perspectives of AI security. First, this project will create a practice-in-the-loop learning experience for students to understand the security of ML in computer vision, such as image recognition and object detection. Educational modules will be developed to cover various ML models for vision sensing and their security vulnerabilities and solutions. Second, this project will extend the interactive learning experience for students to understand the security problems of ML in voice assistant systems, such as speech recognition and speaker identification. The educational modules will be developed to introduce ML models for audio data processing and security vulnerabilities in voice assistant AI systems. Third, this project will develop software-based labs and training projects to enhance students’ understanding. The outcomes of this project, such as teaching slides, software labs, and training projects, will enable various undergraduate student training and outreach activities. They will also be disseminated online and through academic publications, ensuring diverse communities can readily access and employ the educational resources. 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

This project evaluates new and old electrification systems to understand outages and rising costs of maintaining and producing electricity. This project seeks to 1) understand how communities navigate changing service conditions, 2) establish theoretical and empirical knowledge on the normalization of infrastructure services, and 3) examine how assumptions in energy studies affect technological choices. Public engagement via forums, books, websites, visual arts, and film will disseminate research findings to a wide audience. This project seeks to understand the evolution and differences between small and large grid systems. It focuses on how everyday experiences of energy networks are instrumental in infrastructure development. It will explore responses to electrical disruptions during times of network expansion and generate new insights for energy infrastructure studies. The project also provides pedagogical materials for understanding how large-scale technological systems change over time. 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

Almost all the mass of atoms resides in the nucleons (protons and neutrons) that make up the corresponding atomic nucleus. Nucleons are not elementary particles but rather complicated bound states of quarks and gluons (collectively denoted as partons), whose interaction is described by quantum chromodynamics (QCD). The parton structure of the nucleon is encoded in different parton distributions and quark-gluon correlations. Both parton distributions and quark-gluon correlations can, in principle, be studied through high-energy (hard) scattering processes at particle accelerators, including the future Electron-Ion Collider in the USA. However, for quark-gluon correlations this is very difficult, and therefore these quantities are presently largely unexplored. The first goal of the project is to develop and apply an improved theoretical framework for extracting the nucleon transversity, one of the least explored parton distributions, from data on high-energy processes. Along with that goes a state-of-the-art extraction of the tensor charge, one of the fundamental properties of the nucleon. The second goal is to develop a formalism which allows for the calculation of quark-gluon correlations numerically in QCD, thus circumventing the lack of information from experiment. The project integrates research and teaching of graduate and undergraduate students, as well as educational outreach. The work on the transversity parton distributions and the tensor charge extends our previous studies in this area. A special focus is on enhancing the QCD formalism for the analysis of data on di-hadron production by computing next-to-leading-order perturbative corrections. This framework then serves as a key ingredient of a novel extraction of the transversity and tensor charge of the nucleon, which uses all pertinent data on di-hadron (and single-hadron) production. A major goal of this study is the comparison of the nucleon tensor charge based on experimental data with benchmark results obtained from first-principles calculations in lattice-QCD. The study of the quark-gluon correlations exploits a new class of Euclidean parton correlators that can be computed in lattice-QCD. While these correlators do not directly provide the QCGs of interest, the two types of quantities can be related through calculations in perturbative QCD. Developing the required formalism, which presently is in its infancies, is a key objective of the project. Close interaction is foreseen with researchers performing numerical calculations in lattice-QCD, to ensure that the new theoretical framework can be readily put into practice. 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 NSF CSforAll Research Practice Partnership (RPP) program aims to provide all U.S. students with the opportunity to participate in computer science and computational thinking education in their schools at the preK-12 levels. STARS Computing Corps, a CISE Broadening Participation in Computing Alliance, in partnership with Trivium Consulting Group, will provide resources and technical support to potential Principle Investigators (PIs) in order to increase their interest and capacity for creating and submitting competitive CSforAll RPP strand grant proposals. By building connections and potential collaborations among researchers and practitioners in diverse regions with diverse backgrounds and experiences, the project has the potential to foster community and introduce new and future PIs to the broader research community. The 2024 CISE CSforAll Technical Support workshops aim to increase the breadth of projects supported through the NSF CSforAll program. The goal will be realized through multi-layered outreach to researchers and/or practitioners whose past experiences, interests, and expertise make them qualified candidates to carry out successful CSforAll proposals; registration and assessment of participant project ideas, teams, and expertise; provision of diverse online workshops and sessions (that can be engaged in synchronously or asynchronously) in critical elements of successful proposal development; and one-on-one technical support in key proposal development activities. 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

Public safety has become an increasingly important issue in the United States due to the potential threat posed by hidden weapons and homemade bombs in public places where extensive security checks are not available. Traditional security systems, such as X-ray machine screening, are expensive and primarily deployed in high-security areas like airports and government buildings. This project proposes to leverage the prevalent WiFi infrastructure in many public spaces to enable hidden object detection. The project team utilizes extracted WiFi signal features to identify the shape of hidden objects and determine their materials, and subsequently be able to detect suspicious items. The success of this project will greatly enhance public safety by offering easy to deploy and low cost detection systems at public venues (e.g., schools, theme parks, and sports stadiums), thereby addressing the urgent need for better safety in everyday public spaces. Building upon the team's previous foundational work, this project investigates using received WiFi signal features to determine the types of materials of objects inside bags. Target identification models and domain adaptation frameworks based on deep learning techniques are designed to ensure a good identification accuracy in diverse environments. Robust shape reconstruction algorithm helps to recognize suspicious objects. Additionally, new mechanisms using directional antennas are developed to mitigate the impact of the bag carrier's movements. The TTP project will create a prototype system and validate the system's functionality, accuracy, and robustness. The project team seeks to integrate the project's research efforts with educational activities such as developing graduate and undergraduate curricula. The team will also recruit underrepresented students into the project. The team will work closely with technology collaborators for field trial and potential deployment into an operational environment. 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 ease of looking up the answers to questions designed to test factual knowledge threatens measures of knowledge and beliefs in all areas of research in the social and natural sciences. This research project will advance methods of detecting and deterring respondents from looking up the answers ("cheating") in online surveys. It will introduce new methods of assessing how cheating affects measures of knowledge and knowledge disparities between social groups. These methods will be tested across a range of substantive areas, including political knowledge, scientific knowledge, health literacy, and beliefs about social and economic conditions. Depending on the prevalence of cheating, the availability and accuracy of information, and respondents' motivations for cheating, the optimal approach to dealing with cheating may vary. The results of this research will help researchers think about whether cheating is likely to be a problem in their substantive area. More generally, the research will enable more accurate measures of knowledge, beliefs, and cognitive traits. The project will produce a series of online resources and tutorials, open access articles, and an open access book. Code and data generated from the project will be publicly available. This research project will produce a broadly applicable set of methodological tools for dealing with cheating in online surveys, as well as a theoretical framework that helps researchers' reason about whether cheating is likely to be a significant threat in their substantive area. First, the project will identify high-performing strategies for detection and deterrence that can be widely implemented without compromising privacy. Detection methods include indicators of page-switching, question timers, and specialized "catch" questions that are unlikely to be answered correctly without cheating. Deterrence methods include instructions and pledges not to look up the answers. Based on a classification framework, the performance of different implementations of the individual methods will be compared, as well as the benefits of combining multiple strategies. Second, the project will conduct a needs assessment to examine how the prevalence of cheating varies across substantive topics and how this affects the optimal combination of strategies for each topic. Finally, the project will introduce new methods of estimating the bias that cheating introduces into measures of knowledge and knowledge disparities between social groups. The theoretical foundation for these tools will shed light on the factors that determine how cheating affects (or does not affect) measured knowledge, including the availability and accuracy of information about the topic, as well as the extent to which lookups are motivated by ignorance or to confirm what one already knows. 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.