Spatiotemporal Learning for Urban Mobility and Exascale Climate Emulation
Title:
Spatiotemporal Learning for Urban Mobility and Exascale Climate Emulation
Abstract:
The proliferation of sensing modalities, together with advances in machine learning and high-performance computing, has opened new methodological questions at the intersection of signal processing and large-scale modelling of complex physical and social systems. In this talk, I will present recent work spanning urban mobility and climate modelling. I will first discuss STEF-DHNet, a spatiotemporal model for ride-hailing demand that combines convolutional and recurrent components with exogenous signals, and the Variational Mode Graph Convolutional Network, which couples variational mode decomposition with graph neural networks for interpretable traffic forecasting. I will then turn to our exascale climate emulator, which combines spherical harmonic transforms with mixed-precision GPU computation to achieve high-fidelity climate modelling at unprecedented scale. I will close with reflections on the methodological themes that recur across these problems, including interpretability, multi-scale decomposition, and the trade-offs between physical structure and learned representations.
Bio:
Dr. Zubair Khalid is an Associate Professor in the Department of Electrical Engineering and the Director of the Centre for Urban Informatics, Technology, and Policy (CITY) at the Lahore University of Management Sciences (LUMS). His research sits at the intersection of artificial intelligence, climate science, and sustainable urban development, and centres on co developing decision support systems with city governments, multilateral organisations, and academic institutions to address challenges in urban mobility, air quality, and climate resilience. His broader interests include machine learning, computer vision, remote sensing, and advanced signal processing, with applications across cosmology, medical imaging, and urban systems. He received his Ph.D. from the Australian National University, is a Senior Member of IEEE, has served as an Associate Editor for IEEE Signal Processing Letters, and is a recipient of the 2024 ACM Gordon Bell Prize. Further information is available at www.zubairkhalid.org and http://city.lums.edu.pk and ANU researchers win Gordon Bell Prize for record-breaking algorithm | Research
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
Loading virtual attendance info...
- The Australian National University
- 35 Science Road
- Acton, Australian Capital Territory
- Australia 2601
- Building: Birch Building - Seminar Room
- Room Number: 2.04
- Click here for Map
- Starts 26 June 2026 02:00 PM UTC
- Ends 23 July 2026 02:00 PM UTC
- No Admission Charge