Research map / Federated learning and privacy / Privacy attacks and secure learning
Inferring Nighttime Satellite Imagery from Human Mobility
Brian Dickinson, Gourab Ghoshal, Xerxes Dotiwalla, Adam Sadilek, Henry Kautz
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 |
|---|---|---|---|
| Replace real VIIRS light with the Week-20 mobility-inferred map | GDP correlation analysis · United States states (All) | Spearman correlation with GDP: 0.8546^{***} -> 0.9639^{***} | Improved |
| Use the All-Weeks inferred map instead of real VIIRS light | GDP correlation analysis · Well-covered countries (All) | Spearman correlation with GDP: 0.9384^{***} -> 0.8420^{***} | Hurt |
| Use Week-20 mobility data rather than All-Weeks data (ablation) | Mobility-inferred light maps · United States states (All) | Spearman correlation with GDP: 0.9598^{***} -> 0.9639^{***} | Improved |
| Restrict All-Weeks mobility data to Spring (ablation) | Mobility-inferred light maps · United States states (All) | Spearman correlation with GDP: 0.9598^{***} -> 0.9584^{***} | No clear effect |
| Restrict All-Weeks mobility data to May (ablation) | Mobility-inferred light maps · United States states (All) | Spearman correlation with GDP: 0.9598^{***} -> 0.9591 | No clear effect |
| Use only Annual mobility data instead of All-Weeks data (ablation) | Mobility-inferred light maps · United States states (All) | Spearman correlation with GDP: 0.9598^{***} -> 0.9375^{***} | Hurt |
| Use Annual rather than All-Weeks mobility data (ablation) | Mobility-inferred light maps · Well-covered countries (All) | Spearman correlation with GDP: 0.8420^{***} -> 0.8738^{***} | Improved |
| Fit separate regional models rather than one global model | Light prediction · Global 2016 light prediction | MAE: 178.47 -> 7.43 | Improved |
| Replace linear regression with random forest regression (ablation) | Global light prediction · Global 2016 light prediction | MAE: 178.47 -> 211.46 | Hurt |
| Use regression-kriging rather than a global linear model | Light prediction · Global 2016 light prediction | MAE: 178.47 -> 7.43 | Improved |
| Use a regional random forest rather than a regional linear model (ablation) | North America light prediction · North America | MAE: 23.70 -> 18.63 | Improved |
| Use a regional random forest rather than a regional linear model (ablation) | Southeastern Asia light prediction · Southeastern Asia | MAE: 108.15 -> 108.00 | Improved |
Closest papers by meaning
- Satellites Reveal Mobility: A Commuting Origin-destination Flow Generator for Global Cities (NeurIPS 2025 · similarity 0.85)
- SynMob: Creating High-Fidelity Synthetic GPS Trajectory Dataset for Urban Mobility Analysis (NeurIPS 2023 · similarity 0.82)
- TraXion: Rethinking Pre-training Frameworks for Mobility and Beyond (NeurIPS 2026 · similarity 0.80)
- CityLens: Evaluating Large Vision-Language Models for Urban Socioeconomic Sensing (ICLR 2026 · similarity 0.80)
- Measuring Fine-Grained Urban Air Temperature with Satellite Imagery (AAAI 2025 · similarity 0.80)
- SeoulMMOD: A Large-Scale Multimodal Origin-Destination Flow Benchmark (NeurIPS 2026 · similarity 0.79)
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.
