Research map / Probabilistic and scientific modeling / Dynamical systems
A Gravity-Informed Spatiotemporal Transformer for Human Activity Intensity Prediction
Yi Wang, Zhenghong Wang, Fan Zhang, Chaogui Kang, Sijie Ruan, Di Zhu, Chengling Tang, Zhongfu Ma, Weiyu Zhang, Yu Zheng, Philip S. Yu, Yu Liu
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 AdaGravity (ablation) | Gravityformer · Minneapolis | RMSE: 29.98 -> 26.45 | Improved |
| add AdaGravity (ablation) | Gravityformer · Boston | RMSE: 18.45 -> 16.90 | Improved |
| add AdaGravity (ablation) | Gravityformer · Chicago | RMSE: 27.43 -> 23.76 | Improved |
| add inflow and outflow features to the mass parameters (ablation) | Gravityformer · Minneapolis | RMSE: 29.11 -> 26.45 | Improved |
| add inflow and outflow features to the mass parameters (ablation) | Gravityformer · Chicago | RMSE: 25.42 -> 23.76 | Improved |
| add adaptive distance scaling (ablation) | Gravityformer · Minneapolis | RMSE: 28.16 -> 26.45 | Improved |
| use distance matrix (distance decay) instead of adaptive adjacency matrix (ablation) | Gravityformer · Minneapolis | RMSE: 29.23 -> 26.45 | Improved |
| add Hadamard mapper (ablation) | Gravityformer · Minneapolis | RMSE: 29.05 -> 26.45 | Improved |
| add dynamic adjustment branches (Conv2former structure) (ablation) | Gravityformer · Minneapolis | RMSE: 27.52 -> 26.45 | Improved |
| use parallel spatiotemporal structure instead of sequential (ablation) | Gravityformer · Minneapolis | RMSE: 26.74 -> 26.45 | Improved |
| make AdaGravity parameters dynamic (ablation) | Gravityformer · Minneapolis vs Boston | RMSE: Minneapolis 30.50 vs 26.45; Boston 16.68 vs 16.90 | Mixed |
| use learnable pooling for mass parameters (ablation) | Gravityformer · Minneapolis | RMSE: 32.93 vs 26.45 | Hurt |
Closest papers by meaning
- Convolutional Tensor-Train LSTM for Spatio-Temporal Learning (NeurIPS 2020 · similarity 0.86)
- Spatiotemporal-Untrammelled Mixture of Experts for Multi-Person Motion Prediction (AAAI 2026 · similarity 0.84)
- UniSTD: Towards Unified Spatio-Temporal Learning across Diverse Disciplines (CVPR 2025 · similarity 0.83)
- GeoTransolver: Learning Physics on Irregular Domains using Multi-scale Geometry Aware Physics Attention Transformer (NeurIPS 2026 · similarity 0.83)
- Earthfarsser: Versatile Spatio-Temporal Dynamical Systems Modeling in One Model (AAAI 2024 · similarity 0.83)
- Wavelet-Driven Spatiotemporal Predictive Learning: Bridging Frequency and Time Variations (AAAI 2024 · similarity 0.82)
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.
