Tag
The paper investigates using graph neural networks to correct spectral prior estimates for connectivity-loss in road networks under disruptions, with evaluations on synthetic and real-world OpenStreetMap datasets.