This article proposes the first refined dataset, A Fine-Grained Global Terraced Parcel and Boundary Dataset(GTPBD), covering typical terraced field distribution areas worldwide. It is innovative and supports multi task evaluation.

Sequential updates with Parallel aggregation for Enhanced Federated Learning under Category and Domain Shifts

We propose DiffLiG, a graph-based model that addresses station heterogeneity and oversmoothing via spatiotemporal encoding and diffusion-based ensemble refinement for accurate, structure-aware precipitation interpolation.