Research map / Label-efficient and robust learning / Recognition and long-tail
Exploiting Pre-trained Language Model for Cross-city Urban Flow Prediction Guided by Information-theoretic Analysis
Qiang Zhou, Xudong Tong, Yuting Liu, ChuanXing Liu, Jingjing Gu
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 pre-trained language model (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 9.033 -> 7.761 | Improved |
| add trend-coefficient branch (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.223 -> 7.761 | Improved |
| add period-coefficient branch (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.073 -> 7.761 | Improved |
| add semantic bridge encoder (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.215 -> 7.761 | Improved |
| add task-specific adapter (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.013 -> 7.761 | Improved |
| add GDAConv in all components (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.226 -> 7.761 | Improved |
| add GDAConv in task adapter (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 8.231 -> 7.761 | Improved |
| add GDAConv in regression layer (ablation) | PLM-CUP-Q · NJ->SH transfer | MAE: 7.979 -> 7.761 | Improved |
| add source-city pretraining and transfer instead of target-only training | PLM-CUP-Q · NJ->SH vs SH target only | MAE: 8.183 -> 7.761 | Improved |
| raise MMD coefficient kappa from 0.01 to 0.1 (ablation) | PLM-CUP-Q · SH target | MAE: MAE rises 7.2% | Hurt |
| lower MMD coefficient kappa from 0.01 to 0.001 (ablation) | PLM-CUP-Q · SH target | MAE: 2.4% drop | Hurt |
| use 1 decomposition iteration instead of 3 (ablation) | PLM-CUP-Q · SH target | MAE: 8.438/8.462 (+8.7%) | Hurt |
Closest papers by meaning
- Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting (AAAI 2025 · similarity 0.87)
- Diffusion Transformers as Open-World Spatiotemporal Foundation Models (NeurIPS 2025 · similarity 0.85)
- GeoPro-Net: Learning Interpretable Spatiotemporal Prediction Models Through Statistically-Guided Geo-Prototyping (AAAI 2025 · similarity 0.85)
- Explainable Origin-Destination Crowd Flow Interpolation via Variational Multi-Modal Recurrent Graph Auto-Encoder (AAAI 2024 · similarity 0.85)
- Spatio-Temporal Few-Shot Learning via Diffusive Neural Network Generation (ICLR 2024 · similarity 0.85)
- STD-PLM: Understanding Both Spatial and Temporal Properties of Spatial-Temporal Data with PLM (AAAI 2025 · similarity 0.84)
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.
