SEDAN: A Diffusion Model for Cross-City Commuting Flow Generation
Researchers propose SEDAN, a Structure-Enhanced Diffusion model conditioned on Attributed Nodes, to generate origin-destination (OD) matrices that generalize across cities. The model treats a city as an attributed graph where regions are nodes with demographic and point-of-interest features, and commuting flows are weighted edges. Adjacency and distance matrices capture spatial structure. A fusion mechanism jointly models semantic and spatial information. The approach addresses heterogeneity in commuting patterns caused by individual intentions, geographic constraints, and social dynamics, aiming to improve urban governance, traffic planning, and resource allocation. The paper is available on arXiv under ID 2605.00938.
Key facts
- SEDAN stands for Structure-Enhanced Diffusion model conditioned on Attributed Nodes.
- The model generates OD matrices that generalize across cities.
- Each region is a node with demographic and POI features.
- Commuting flows are modeled as weighted edges.
- Adjacency and distance matrices characterize spatial structure.
- A fusion mechanism jointly models semantic and spatial information.
- The approach addresses heterogeneity in commuting patterns.
- Paper available on arXiv: 2605.00938.
Entities
Institutions
- arXiv