Robust Space-Time Modeling with Adaptive AI
- Event Type
- Seminar/Symposium
- Sponsor
- Geography & GIS
- Location
- Room 2049 Natural History Building (and via Zoom)
- Speaker
- Dr. Zhaonan Wang, Department of Geography & GIS and I-GUIDE
- Cost
- This talk is free and open to the public.
- Registration
- Join via Zoom
- Contact
- Geography & GIS
- geography@illinois.edu
- Views
- 144
- Originating Calendar
- Geography and Geographic Information Science
Rapidly developing mobile, social, and sensor networks are accumulating massive volumes of geospatial and temporal data. Space-time modeling on these data is a fundamental problem in building decision support systems for applications like traffic management. In a real-world environment, such spatio-temporal data show high heterogeneity over space and non-stationarity over time, which makes the prediction task especially challenging.
My research focuses on enhancing the robustness of space-time models utilizing adaptive AI techniques, such as meta learning. Making robust predictions lays a foundation for not only inclusive spatial planning, but also emergency response to adverse events, including traffic accidents, COVID pandemic, and natural disasters. The robust space-time models will contribute to a more inclusive and adaptive society in a changing environment, which aligns well with the Sustainable Development Goals by the United Nations.
Dr. Zhaonan Wang is a GIScientist whose research focuses on geospatial, spatio-temporal data and AI-driven decision making for urban and societal problems. He completed his PhD in Spatial Information Science at the University of Tokyo and is currently a postdoc in GGIS, working with Dr. Shaowen Wang on geographic information retrieval and language models.