Meiliu and Fengjiao at COSIT 2026 in York, 22–25 September
Fengjiao Li and I attended the 17th Conference on Spatial Information Theory (COSIT 2026) in York, UK, from 22–25 September. The conference theme, Cognition and Flows, brought together work on how people and systems experience, represent and reason about space.
Fengjiao presented our short paper, “Spatiotemporal Adaptive Learning for Flow-Based Graph Neural Networks,” by Fengjiao Li, Meiliu Wu and Ana Basiri. The paper is published open access in the COSIT 2026 proceedings, Leibniz International Proceedings in Informatics (LIPIcs, volume 393, pages 31:1–31:9). Read the paper.
Our study learns how much different spatial relationships—geographic adjacency, road connectivity and mobility links—matter when forecasting COVID-19 dynamics in Glasgow and Edinburgh. It also learns how to weight different forecasting horizons, making the model’s spatial and temporal choices explicit.
It was a pleasure to share the conference with Fengjiao and take part in thoughtful conversations across spatial information theory and GeoAI. Thank you to the COSIT 2026 organising committee and session chairs for a welcoming conference and an engaging programme.
Organising committee
- General Chair: Ed Manley (University of Leeds)
- Local Organising Committee: Gabriele Filomena (University of Liverpool), Lex Comber and Ed Manley (University of Leeds)
- Programme Chairs: Sabine Timpf (University of Augsburg), Gabriele Filomena (University of Liverpool), Rui Zhu (University of Bristol), Nicholas Giudice (University of Maine) and Armand Kapaj (University of Zurich)
- Themed Sessions Chair: Clio Andris (Georgia Institute of Technology)
- Doctoral Mentorship Chair: Crystal Bae (University of Chicago)
- Communications Chair: Armand Kapaj (University of Zurich)
Full COSIT 2026 organising committee
At COSIT 2026
Presenting our paper
