Research map / Video / Video segmentation and tracking
UrbanPG: An Efficient Framework with Personalized Context and General Backbone Interaction for Urban Spatio-Temporal Learning
Aoyu Liu, Yaying Zhang
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 temporal context prompt (ablation) | UrbanPG · SD / CA | MAE, RMSE, MAPE | Improved |
| add spatial context prompt (ablation) | UrbanPG · SD / CA | MAE, RMSE, MAPE | Improved |
| add random perturbation regularization on spatial embeddings (ablation) | UrbanPG · SD / CA | MAE, RMSE, MAPE | Improved |
| add STCA linear attention module (ablation) | UrbanPG · SD / CA | MAE, RMSE, MAPE | Improved |
| set perturbation ratio p to 0.1 (vs higher) (ablation) | UrbanPG · SD / CA | MAE | Improved |
| set feature dimension d to 256 (ablation) | UrbanPG · SD / CA | MAE | Mixed |
| use UrbanPG instead of PatchSTG | UrbanPG vs PatchSTG · LargeST CA | training time / inference time / memory: -48.96% / -72.44% / -45.72% | Improved |
| UrbanPG vs PatchSTG | UrbanPG vs PatchSTG · LargeST GBA | MAPE: 14.64% -> 14.61% | No clear effect |
| freeze backbone and fine-tune prompts after pre-training vs FlashST | UrbanPG · LargeST CA-D3 (few-shot) | MAE: 18.91 -> 18.28 | Improved |
| expand spatial prompts with frozen backbone vs EAC | UrbanPG · PEMS-Stream (continual) | MAE: 13.49 -> 10.77 | Improved |
Closest papers by meaning
- Get Rid of Isolation: A Continuous Multi-task Spatio-Temporal Learning Framework (NeurIPS 2024 · similarity 0.90)
- Diffusion Transformers as Open-World Spatiotemporal Foundation Models (NeurIPS 2025 · similarity 0.89)
- Geolocation Representation from Large Language Models Are Generic Enhancers for Spatio-Temporal Learning (AAAI 2025 · similarity 0.87)
- GeoPro-Net: Learning Interpretable Spatiotemporal Prediction Models Through Statistically-Guided Geo-Prototyping (AAAI 2025 · similarity 0.87)
- Spatio-Temporal Few-Shot Learning via Diffusive Neural Network Generation (ICLR 2024 · similarity 0.87)
- Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting (AAAI 2025 · similarity 0.86)
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.
