New dataset and benchmark aim to advance air quality forecasting with foundation models

Researchers have introduced the Air Quality Arena (AQA), a comprehensive dataset and benchmark designed to improve air quality forecasting using time-series foundation models (TSFMs). AQA-Data encompasses six major pollutants across 7,000+ stations in 7 countries over three years, while AQA-Bench evaluates model performance. Initial results show TSFMs significantly outperform traditional methods, with a cross-modal architecture combining vision and time-series models achieving the best zero-shot forecasting accuracy. AI

IMPACT This new dataset and benchmark could accelerate the development and adoption of advanced AI models for critical environmental monitoring and public health applications.

RANK_REASON The item describes a new dataset and benchmark for evaluating time-series foundation models on air quality forecasting, published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Air Quality Arena
• alphaXiv
• AQA-Bench
• AQA-Data
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• Pandarasamy Arjunan
• ScienceCast

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →