{"ID":2879333,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.16161","arxiv_id":"2508.16161","title":"STA-GANN: A Valid and Generalizable Spatio-Temporal Kriging Approach","abstract":"Spatio-temporal tasks often encounter incomplete data arising from missing or inaccessible sensors, making spatio-temporal kriging crucial for inferring the completely missing temporal information. However, current models struggle with ensuring the validity and generalizability of inferred spatio-temporal patterns, especially in capturing dynamic spatial dependencies and temporal shifts, and optimizing the generalizability of unknown sensors. To overcome these limitations, we propose Spatio-Temporal Aware Graph Adversarial Neural Network (STA-GANN), a novel GNN-based kriging framework that improves spatio-temporal pattern validity and generalization. STA-GANN integrates (i) Decoupled Phase Module that senses and adjusts for timestamp shifts. (ii) Dynamic Data-Driven Metadata Graph Modeling to update spatial relationships using temporal data and metadata; (iii) An adversarial transfer learning strategy to ensure generalizability. Extensive validation across nine datasets from four fields and theoretical evidence both demonstrate the superior performance of STA-GANN.","short_abstract":"Spatio-temporal tasks often encounter incomplete data arising from missing or inaccessible sensors, making spatio-temporal kriging crucial for inferring the completely missing temporal information. However, current models struggle with ensuring the validity and generalizability of inferred spatio-temporal patterns, esp...","url_abs":"https://arxiv.org/abs/2508.16161","url_pdf":"https://arxiv.org/pdf/2508.16161v2","authors":"[\"Yujie Li\",\"Zezhi Shao\",\"Chengqing Yu\",\"Tangwen Qian\",\"Zhao Zhang\",\"Yifan Du\",\"Shaoming He\",\"Fei Wang\",\"Yongjun Xu\"]","published":"2025-08-22T07:33:12Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\"]","methods":"[\"Generative Adversarial Network\",\"Graph Neural Network\"]","has_code":false}