AnchorGK: Anchor-based Incremental and Stratified Graph Learning Framework for Inductive Spatio-Temporal Kriging
Xiaobin Ren, Kaiqi Zhao, Katerina Taškova, Patricia Riddle
TL;DR
AnchorGK tackles inductive spatio-temporal kriging under sparse sensor deployment and incomplete feature availability by introducing anchor-based stratification and a dual-view graph learning layer. It creates fine-grained strata around anchor locations and models spatial correlations with a density-aware adjacency, while CFE and CSE fuse cross-feature and cross-strata information in a Mixture-of-Experts framework, enabling inductive inference for unseen locations. Local/global kriging provides initial estimates, which are refined through a Kalman-filter-based fusion and iterative graph updates during training. Empirical results on three real-world datasets show AnchorGK consistently outperforms state-of-the-art baselines, particularly under high sparsity and feature missingness, highlighting its ability to capture spatial heterogeneity and leverage heterogeneous feature availability for accurate spatio-temporal kriging.
Abstract
Spatio-temporal kriging is a fundamental problem in sensor networks, driven by the sparsity of deployed sensors and the resulting missing observations. Although recent approaches model spatial and temporal correlations, they often under-exploit two practical characteristics of real deployments: the sparse spatial distribution of locations and the heterogeneous availability of auxiliary features across locations. To address these challenges, we propose AnchorGK, an Anchor-based Incremental and Stratified Graph Learning framework for inductive spatio-temporal kriging. AnchorGK introduces anchor locations to stratify the data in a principled manner. Anchors are constructed according to feature availability, and strata are then formed around these anchors. This stratification serves two complementary roles. First, it explicitly represents and continuously updates correlations between unobserved regions and surrounding observed locations within a graph learning framework. Second, it enables the systematic use of all available features across strata via an incremental representation mechanism, mitigating feature incompleteness without discarding informative signals. Building on the stratified structure, we design a dual-view graph learning layer that jointly aggregates feature-relevant and location-relevant information, learning stratum-specific representations that support accurate inference under inductive settings. Extensive experiments on multiple benchmark datasets demonstrate that AnchorGK consistently outperforms state-of-the-art baselines for spatio-temporal kriging.