Research map / Architectures and efficiency / Attention and transformers
AirFormer: Predicting Nationwide Air Quality in China with Transformers
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use AirFormer instead of DeepAir | PM2.5 forecasting · 1-24h | MAE: 17.47 -> 16.03 | Helped |
| use AirFormer instead of MTGNN | PM2.5 forecasting · 49-72h | RMSE: 55.73 -> 50.22 | Helped |
| add DS-MSA spatial modeling (ablation) | AirFormer · PM2.5, 1-24h | MAE: 17.24 -> 16.03 | Helped |
| replace standard MSA with DS-MSA (ablation) | AirFormer · PM2.5, 1-24h | MAE: 18.19 -> 16.03 | Helped |
| replace local MSA with DS-MSA (ablation) | AirFormer · PM2.5, 1-24h | MAE: 18.63 -> 16.03 | Helped |
| use a 50km-and-200km dartboard instead of 50km only (ablation) | AirFormer DS-MSA · PM2.5 forecasting | MAE: 16.43 -> 16.03; 22.87 -> 21.65; 24.32 -> 23.64 | Helped |
| add a 500km dartboard radius to 50-200 (ablation) | AirFormer DS-MSA · PM2.5 forecasting | MAE: 16.03 -> 16.09; 21.65 -> 21.30; 23.64 -> 22.85 | Mixed |
| use the 50-200 dartboard instead of 50-200-500 (ablation) | AirFormer DS-MSA · training epoch time | time/epoch: 2,285 -> 1,708 | Helped |
| add CT-MSA temporal modeling (ablation) | AirFormer · PM2.5 forecasting | MAE | Helped |
| replace WaveNet with MSA-based temporal modeling (ablation) | AirFormer · PM2.5 forecasting | MAE | Helped |
| add causality and local windows to temporal MSA (ablation) | AirFormer · PM2.5 forecasting | MAE | Helped |
| add stochastic latent variables (ablation) | AirFormer · PM2.5, sudden change | MAE: 57.52 -> 54.92 | Helped |
Closest papers by meaning
- AirRadar: Inferring Nationwide Air Quality in China with Deep Neural Networks (AAAI 2025 · similarity 0.89)
- AirPhyNet: Harnessing Physics-Guided Neural Networks for Air Quality Prediction (ICLR 2024 · similarity 0.86)
- Air Quality Prediction with Physics-Guided Dual Neural ODEs in Open Systems (ICLR 2025 · similarity 0.84)
- On the Integration of Spatial-Temporal Knowledge: A Lightweight Approach to Atmospheric Time Series Forecasting (NeurIPS 2025 · similarity 0.81)
- SNN-PDE: Learning Dynamic PDEs from Data with Simplicial Neural Networks (AAAI 2024 · similarity 0.80)
- UrbanDataLayer: A Unified Data Pipeline for Urban Science (NeurIPS 2024 · similarity 0.80)
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.