aiqbench analysis
MoRA introduces a novel geospatial intelligence framework that treats human mobility graphs as the primary backbone for learning location embeddings, effectively fusing remote sensing data with demographic statistics. By leveraging Graph Neural Networks (GNNs) and asymmetric contrastive learning on a billion-edge graph, the model achieves significant performance gains in predicting socio-economic contexts across diverse domains. This work represents a meaningful advancement in scaling geospatial representation learning by grounding physical location meaning in fundamental human dynamics rather than just static Earth observation data.
MoRA: Mobility as the Backbone for Geospatial Representation Learning at Scale
arXiv:2506.01297v5 Announce Type: replace Abstract: Representation learning of geospatial locations remains a core challenge in achieving general geospatial intelligence, with increasingly diverging philosophies and techniques. While Earth observation paradigms excel at depicting locations in their physical states, we claim that a location's comprehensive "meaning" is better grounded in its internal human activity patterns and, crucially, its functional relationships with other locations, as revealed by human movement. We present MoRA, a human-centric geospatial framework that leverages a mobility graph as its core backbone to fuse various data modalities, aiming to learn embeddings that represent the socio-economic context and functional role of a location. MoRA achieves this through the integration of spatial tokenization, GNNs, and asymmetric contrastive learning to align 100M+ POIs, massive remote sensing imagery, and structured demographic statistics with a billion-edge mobility graph, ensuring the three auxiliary modalities are interpreted through the lens of fundamental human dynamics. To rigorously evaluate the effectiveness of MoRA, we construct a benchmark dataset composed of 9 downstream prediction tasks across social and economic domains. Experiments show that MoRA, with four input modalities and a compact 128-dimensional representation space, achieves superior predictive performances than state-of-the-art models by an average of 12.9%. Echoing LLM scaling laws, we further demonstrate the scaling behavior in geospatial representation learning. We open-source code and pretrained models at: https://github.com/ylzhouchris/MoRA.
Abstract:Representation learning of geospatial locations remains a core challenge in achieving general geospatial intelligence, with increasingly diverging philosophies and techniques. While Earth observation paradigms excel at depicting locations in their physical states, we claim that a location's comprehensive "meaning" is better grounded in its internal human activity patterns and, crucially, its functional relationships with other locations, as revealed by human movement. We present MoRA, a human-centric geospatial framework that leverages a mobility graph as its core backbone to fuse various data modalities, aiming to learn embeddings that represent the socio-economic context and functional role of a location. MoRA achieves this through the integration of spatial tokenization, GNNs, and asymmetric contrastive learning to align 100M+ POIs, massive remote sensing imagery, and structured demographic statistics with a billion-edge mobility graph, ensuring the three auxiliary modalities are interpreted through the lens of fundamental human dynamics. To rigorously evaluate the effectiveness of MoRA, we construct a benchmark dataset composed of 9 downstream prediction tasks across social and economic domains. Experiments show that MoRA, with four input modalities and a compact 128-dimensional representation space, achieves superior predictive performances than state-of-the-art models by an average of 12.9%. Echoing LLM scaling laws, we further demonstrate the scaling behavior in geospatial representation learning. We open-source code and pretrained models at: this https URL.
Artificial Intelligence (cs.AI)
arXiv:2506.01297 [cs.AI]
(or arXiv:2506.01297v5 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2506.01297 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 307 words
Related coverage
Mobility-Embedded POIs: Learning What A Place Is and How It Is Used from Human Movement
arXiv · Jun 9, 2026
Toward a foundation model for forest point clouds
arXiv · Sep 22, 2026
How mobility gives language models a deeper understanding of place
Google Research · Aug 21, 2026
UniWAM Technical Report: Unified Mobile Manipulation via Mixed-Stream World-Action Modeling and Manipulation Anchor Pose Supervision
arXiv · Oct 1, 2026