Research map / Time series / Long-term forecasting
Real-Time Long Horizon Air Quality Forecasting via Group-Relative Policy Optimization
Inha Kang, Eunki Kim, Wonjeong Ryu, Jaeyo Shin, Seungjun Yu, Yoon-Hee Kang, Seongeun Jeong, Eunhye Kim, Soontae Kim, Hyunjung Shim
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 CMAQ reanalysis to OBS training data (ablation) | SFT model · PM2.5 SFT (2023 test) | F1: 50.74 -> 54.40 | Improved |
| add CMAQ reanalysis to OBS training data (ablation) | SFT model · PM10 SFT (2023 test) | F1: 41.63 -> 52.65 | Improved |
| add temporal accumulation loss T=2 (ablation) | SFT model · PM2.5 SFT (2023 test) | F1: 54.40 -> 57.65 | Improved |
| extend temporal accumulation loss from T=2 to T=4 (ablation) | SFT model · PM2.5 SFT (2023 test) | F1: 57.65 -> 59.90 | Improved |
| add temporal accumulation loss (T=4 vs none) (ablation) | SFT model · PM10 SFT (2023 test) | F1: 52.65 -> 57.32 | Improved |
| extend TA loss from T=2 to T=4 (ablation) | SFT model · PM10 SFT (2023 test) | F1: 54.60 -> 57.32 | Improved |
| fine-tune Aurora regionally on OBS only | Aurora · PM2.5 (2023 test) | F1: 16.06 -> 50.74 | Improved |
| apply GRPO with AQI reward after SFT | FAKER-Air SFT · PM2.5 (2023 test, 120h) | False Alarm Rate: 32.86 -> 17.32 | Improved |
| apply GRPO with AQI reward after SFT | FAKER-Air SFT · PM10 (2023 test, 120h) | False Alarm Rate: 18.44 -> 10.81 | Improved |
| apply GRPO after SFT | FAKER-Air SFT · PM2.5 (2023 test, 120h) | Bias: 1.52 -> 0.96 | Improved |
| use AQI class-wise reward instead of MSE reward (ablation) | FAKER-Air GRPO · PM2.5 binary F1 (2023 test) | F1: 48.44 -> 56.28 | Improved |
| use AQI class-wise reward instead of MSE reward (ablation) | FAKER-Air GRPO · PM2.5 (2023 test) | False Alarm Rate: 10.54 -> 24.19 | Hurt |
Closest papers by meaning
- AirPhyNet: Harnessing Physics-Guided Neural Networks for Air Quality Prediction (ICLR 2024 · similarity 0.83)
- Air Quality Prediction with Physics-Guided Dual Neural ODEs in Open Systems (ICLR 2025 · similarity 0.83)
- AirRadar: Inferring Nationwide Air Quality in China with Deep Neural Networks (AAAI 2025 · similarity 0.82)
- Quantile-Regression-Ensemble: A Deep Learning Algorithm for Downscaling Extreme Precipitation (AAAI 2024 · similarity 0.82)
- SPATIALEPIBENCH: Benchmarking Spatial Information and Epidemic Priors in Forecasting (NeurIPS 2026 · similarity 0.82)
- DAWP: A framework for global observation forecasting via Data Assimilation and Weather Prediction in satellite observation space (NeurIPS 2025 · 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.
