Back to feed
arXiv cs.LG
arXiv cs.LG
7/15/2026
OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes

OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes

Short summary

OmniPMNet is a ConvCNP-based fusion model that bridges station-level GNN forecasts and gridded CAMS forecasts for PM10 prediction over a 108-hour horizon. Evaluated across 1,618 stations in China throughout 2024, it matches GNN station accuracy (21.14 vs 22.00 µg/m³ MAE) while reducing CAMS gridded error by 30%. Its strongest gains come during dust episodes and high-concentration tails, improving both detection skill and spatial trajectory tracking.

  • •Fusion model combining GNN station forecasts with CAMS gridded forecasts via ConvCNP
  • •30% MAE reduction over CAMS while matching GNN station-level accuracy
  • •Largest gains in dust-storm episodes and high-concentration tails

Generated with AI, which can make mistakes.

#research-breakthrough#ai-tools
Read full article at arXiv cs.LG

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

TSSM: Triaxial State Space Model for Global Station Weather Forecasting with Temporal-Variable-Historical Modeling
arXiv cs.LG
STKAN: Kolmogorov-Arnold Networks for Spatio-Temporal Forecasting
arXiv cs.LG
Omni-Sleep: A Sleep Foundation Model via Hierarchical Contrastive Learning of CNS--ANS Dynamic
arXiv cs.LG
STAGformer: A Spatio-temporal Agent Graph Transformer for Micro Mobility Demand Forecasting
arXiv cs.LG