Georgia Tech Research Corporation

NSF Awards · FY 2024 · 2024-01

Various emerging fields, such as connected autonomous vehicles and smart home analytics, have ushered in an era of Artificial Intelligence (AI) on Internet of Things devices. However, with the proliferation of modern edge devices characterized by limited storage, heterogeneous capabilities, dynamic network connection, and growing concerns of data privacy, it will become impractical to scale and update the current mainstream centralized machine learning (ML) models, leading to large latency delays, energy dissipation, and potentially outdated models with degraded performance. Hence, it has become paramount to efficiently process the inherently decentralized data streams on-device, i.e., closest to their sources without sharing and accumulating the raw training samples on a centralized server. While Federated learning (FL) has emerged to bring ML models, its global model updated at the server by aggregating local models can lead to poor model convergence and requires compromises between model accuracy and available resources on heterogeneous resource-constrained edge devices. Moreover, FL has not yet been designed for handling streaming data generated on the edge. This project aims to open up a new paradigm for developing powerful ML models by combining the best of both worlds of centralized ML and decentralized FL, and thus push forward the frontier of unleashing the great promise of AI to transform human life. The outcomes from this project will lead to new course materials spanning several areas of ML (e.g., decentralized optimization, computer architecture, and edge computing systems) and open-education resources that aim to attract diverse groups of students and eventually deliver a platform for inclusion and innovation. This project is to bridge Centralized ML and decentralized FL, considering the salient streaming and statistical characteristics of data combined with widespread device and network heterogeneity. The key contributions are to develop rigorous foundations for the new decentralized ML training setup in: (1) creating new algorithms for on-device summarization of streaming data to reduce memory cost and improve the processing latency, while preserving privacy; (2) performing decentralized optimization and communication under the heterogeneity of devices in a communication network; and (3) co-designing energy-efficient hardware architecture and algorithm for accelerating streaming data summarization with real-time inference, and developing decentralized incremental learning via streaming data summarization on the edge on a network of real heterogeneous edge devices for system evaluation, validation, and demonstration while promoting green AI. 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 2023 · 2023-10

This research will provide rich insights into a new line of regional experiments with computer-based economic development in the American Midwest, rooted in collaborations between government, industry, and universities to drive their traditions of manufacturing excellence into the next generation. The research will inform the development of sociotechnical interventions supporting bottom-up innovation procedures and emergent outcomes. Such implications will be informed by cutting edge research on data collection and analysis techniques. The project's emphasis on marginalized or excluded stakeholders will offer insights on how to ensure such systems maximize stakeholder agency at all stages of the process. Research outcomes include the identification of transferable practices, policies, and frameworks. These will contribute actionable tactics on how to pursue innovation agendas in contemporary economic contexts characterized by the mobility of information, capital, and talent; emerging forms of work; and contemporary sociotechnical infrastructures. This research will also contribute methods to help researchers better understand stakeholder needs at the regional scale, with a particular focus on populations that have traditionally been underserved. The project will include in-depth ethnographic research and stakeholder-driven engagements with the following two groups: (1) those who are at the forefront of conceiving, designing and implementing techno-urban experiments of fab cities, Internet of Things cities, and smart zones; (2) underserved and excluded populations, most immediately affected by these technological advances. By conducting sociotechnical interventions with community stakeholders, the research advances their commitments to broaden participation. Specifically, this research will build self-efficacy through the clarification of an urban area's own regional advantage. The project will do so through (1) skill building, including building skills with new technologies such as digital fabrication; (2) developing community's capacities to act in a collective way, supporting democracy and self-determination; (3) regional knowledge production and dissemination, including successful community models, actions, and communication tactics; and (4) including diverse actors in spite of their different histories, legacies, aspirations, and concerns, that is, fostering a sense of community in using information technology to build a better future. 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 2023 · 2023-10

This project has directions both in term of advancing our understanding of mathematics and in building the nation's scientific and technical workforce. The mathematical part aims to advance our understanding of the shapes that surfaces present when they are most efficiently navigating their environment. Of course, the notion of efficient depends on the context, so the project considers a number of settings, expecting to find both differences and similarities in the optimal shapes as the criteria for "best shape" are changed. In terms of education, the setting is that nation will need about a million more engineers in the coming decade than we expect the pipeline, as it is currently configured, to produce. At the same time, students from less well-resourced high schools, even if smart and hard-working and interested in a career in science, technology, engineering or mathematics, leave those STEM fields at an alarming rate, as they have trouble transitioning from high school to college. A program led by the PI has achieved notable success in cutting the attrition from STEM students of high potential but less-than-optimal preparation: the grant will help grow, sustain, develop and disseminate information about this comprehensive holistic approach to retention of students in STEM. The project will investigate, via harmonic maps, the asymptotic holonomy of surface group representations in the Hitchin component of several low rank Lie groups. The equivariant harmonic maps from surfaces to the associated symmetric spaces have holomorphic invariants, the geometric topology of which can predict the holonomy of the representation, up to a decaying error. At the same time, the error estimates are strong enough to suggest a unity of approaches: a rescaling of the range and the maps produces a harmonic map to a building, while an apparently different building may be constructed algebraically via an associated real closed field and a valuation. Other projects include finding a new basic minimal surface in three-space through moduli space techniques, a new type of uniformized metric through geometric analytic techniques, and a refinement of a classical circle-packing result on surfaces. The PI will continue his mentorship of undergraduates, graduate students, and postdoctoral scholars. 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.