aiqbench analysis
This research introduces LEON, a self-supervised framework that adapts Graph Masked Autoencoders to learn robust spatial embeddings from OpenStreetMap data using hexagonal grid graphs. By leveraging crowdsourced amenity distributions, the method achieves measurable accuracy gains in downstream tasks such as satellite imagery classification and urban analytics without requiring labeled training data. The study validates self-supervised learning as a viable paradigm for extracting meaningful patterns from complex geospatial datasets.
LEON: Location Embeddings from OSM Neighborhoods via Hexagonal Graph Masked Autoencoders
arXiv:2610.05497v1 Announce Type: new Abstract: Geographic information systems increasingly rely on sophisticated spatial representation learning techniques to extract meaningful patterns from complex geospatial data. This paper introduces LEON, a novel self-supervised framework that adapts Graph Masked Autoencoders (GraphMAE) for geospatial region representation learning. Our method leverages the inherent spatial structure of geographic data by constructing hexagonal grid graphs using H3 indexing and applying masked autoencoding techniques to learn robust spatial embeddings from OpenStreetMap (OSM) amenity distribution patterns. We evaluate LEON on multiple real-world datasets including EuroSAT satellite imagery classification and various geographic prediction tasks (housing prices, crime prediction, and urban analytics). Experimental results demonstrate that LEON achieves significant improvements in spatial understanding, with up to 1.87% accuracy improvement on EuroSAT classification and consistent performance gains across geographic prediction benchmarks. The learned embeddings exhibit highly structured and distinct properties, making them particularly suitable for downstream spatial analysis tasks. Our findings suggest that self-supervised learning provides an effective paradigm for geospatial region representation learning using widely available crowdsourced data.
Abstract:Geographic information systems increasingly rely on sophisticated spatial representation learning techniques to extract meaningful patterns from complex geospatial data. This paper introduces LEON, a novel self-supervised framework that adapts Graph Masked Autoencoders (GraphMAE) for geospatial region representation learning. Our method leverages the inherent spatial structure of geographic data by constructing hexagonal grid graphs using H3 indexing and applying masked autoencoding techniques to learn robust spatial embeddings from OpenStreetMap (OSM) amenity distribution patterns. We evaluate LEON on multiple real-world datasets including EuroSAT satellite imagery classification and various geographic prediction tasks (housing prices, crime prediction, and urban analytics). Experimental results demonstrate that LEON achieves significant improvements in spatial understanding, with up to 1.87% accuracy improvement on EuroSAT classification and consistent performance gains across geographic prediction benchmarks. The learned embeddings exhibit highly structured and distinct properties, making them particularly suitable for downstream spatial analysis tasks. Our findings suggest that self-supervised learning provides an effective paradigm for geospatial region representation learning using widely available crowdsourced data.
15 pages, 5 figures, 5 tables. Pre-review preprint version of a paper published in Advances in Computational Collective Intelligence (ICCCI 2026), CCIS 3044, Springer, pp. 274-287
Machine Learning (cs.LG)
arXiv:2610.05497 [cs.LG]
(or arXiv:2610.05497v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.05497 arXiv-issued DOI via DataCite (pending registration)
Advances in Computational Collective Intelligence, ICCCI 2026, Communications in Computer and Information Science, vol. 3044, Springer, pp. 274-287 (2027)
https://doi.org/10.1007/978-3-032-37936-8_19 DOI(s) linking to related resources
Full text under the publisher's open licence · 231 words
Related coverage
LATTICE: Graph Self-Supervised Learning for Multimodal Spatial Omics Integration
arXiv · Jul 21, 2026
Presentation: Postgres for Production Agents: Your Relational Foundation for Enterprise AI
InfoQ · Jul 15, 2026
GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
Meta Engineering · Aug 3, 2026
Is INT8 Portable? A Cross-Platform Measurement Study of Quantized Inference on Embedded and Automotive Accelerators
arXiv · Sep 16, 2026