Fangzheng Lyu leads NSF project to bring AI and advanced computing to urban research and education
The three-year CITY-AI project will prepare geographers, planners, students, and researchers to use artificial intelligence and high-performance computing to better understand cities.
“At its core, this is a workforce development project,” Lyu said. “We want to train students and practitioners who can use advanced models and computing resources and apply what they learn to urban management and research.” Photo by Chris Moody for Virginia Tech.
Fangzheng Lyu, an assistant professor in Virginia Tech’s Department of Geography, is leading a three-year, nearly $500,000 National Science Foundation project to help urban researchers and practitioners take advantage of artificial intelligence and advanced computing resources.
The project, “CyberTraining: Implementation: Small: CITY-AI: CyberTraining for AI-Enabled Intelligent Urban Informatics and Smart City Innovation,” is supported through the NSF’s Training-based Workforce Development for Advanced Cyberinfrastructure program. The program is led by the Office of Advanced Cyberinfrastructure and works to expand the scientific workforce’s ability to use advanced computing, large-scale data, software, and networking resources.
Lyu is collaborating with co-principal investigators Alexander Michels, an assistant professor at The University of Texas at Dallas, and Carol X. Song of Purdue University’s Rosen Center for Advanced Computing.
Advanced cyberinfrastructure, including supercomputers and systems capable of processing large datasets, is already widely used in fields such as physics and biology. Researchers and practitioners in geography, urban planning, and related social sciences, however, may have less experience with those resources.
“A lot of money has been invested in advanced cyberinfrastructure, but these resources are used less in geography, social science, urban management, and urban planning,” Lyu said. “We want to show people what is possible when they combine these resources with artificial intelligence.”
CITY-AI will create training materials for geographers, urban planners, graduate students, researchers, and practitioners who may have limited backgrounds in artificial intelligence, programming, or high-performance computing.
The materials will introduce participants to advanced cyberinfrastructure, urban big data, artificial intelligence, and applications that integrate those tools to address urban questions. The project team plans to make the materials available online, organize five workshops, offer webinars, and hold a summer school for approximately 20 graduate students.
Participants will learn how computing resources and AI models can support research involving transportation, environmental conditions, urban development, and other aspects of city management.
The project also includes a research component demonstrating how the tools can be used in practice. One application will combine high-performance computing with deep-learning models, vision transformers, and visual-language models to study street-level imagery.
By comparing images of the same location over time, researchers could automatically identify changes such as new buildings, roads, traffic lanes, green infrastructure, renovations, or the conversion of an older building to a new use.
“At its core, this is a workforce development project,” Lyu said. “We want to train students and practitioners who can use advanced models and computing resources and apply what they learn to urban management and research.”
