Research map / Vision-language and multimodal / Human and embodied modeling
MoRA: Mobility as the Backbone for Geospatial Representation Learning at Scale
Ya Wen, Jixuan Cai, Qiyao Ma, Linyan Li, Xinhuan Chen, Chris Webster, Yulun Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add demographics modality (ablation) | MoRA · 9 tasks avg | R2: 0.748 -> 0.779 | Improved |
| add POI modality (ablation) | MoRA · 9 tasks avg | R2: 0.765 -> 0.779 | Improved |
| add satellite imagery modality (ablation) | MoRA · 9 tasks avg | R2: 0.762 -> 0.779 | Improved |
| add CLIP-based alignment (ablation) | MoRA · downstream tasks | R2: 6.5% | Improved |
| use mobility graph instead of neighborhood graph (ablation) | MoRA · downstream tasks | R2: 12.2% | Improved |
| add LINE initialization (ablation) | MoRA · downstream tasks | R2: 1-2% | Improved |
| add mobility graph encoder instead of MLP (ablation) | MoRA · downstream tasks | R2 | Improved |
| use mobility as anchor modality vs demo anchor (ablation) | MoRA · downstream tasks | R2: 16.8% | Improved |
| top-k sampling instead of random sampling (ablation) | MoRA · downstream tasks | R2 | Improved |
| expand pretraining coverage from Jiangsu to China | MoRA · ENE (Jiangsu test) | R2: 0.733 -> 0.889 | Improved |
| expand pretraining coverage from Jiangsu to China | MoRA · HOU (Jiangsu test) | R2: 0.529 -> 0.581 | Mixed |
| finer H3 resolution lv5 to lv7 | MoRA · NTL (East China) | R2: 0.462 -> 0.741 | Improved |
Closest papers by meaning
- TorchSpatial: A Location Encoding Framework and Benchmark for Spatial Representation Learning (NeurIPS 2024 · similarity 0.87)
- GeoCore-9B: Towards Geo-Aware Generative Foundation Models in Earth Observation (NeurIPS 2026 · similarity 0.86)
- SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery (AAAI 2025 · similarity 0.86)
- Towards Faithful Reasoning in Remote Sensing: A Perceptually-Grounded GeoSpatial Chain-of-Thought for Vision-Language Models (ICLR 2026 · similarity 0.86)
- GeoViS: Geospatially Rewarded Visual Search for Remote Sensing Visual Grounding (CVPR 2026 · similarity 0.85)
- Textual Supervision Enhances Geospatial Representations in Vision-Language Models (ICML 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
