Machine learning models benchmarked for atmospheric seeing forecasting

Researchers have evaluated machine learning models for short-term atmospheric seeing forecasting, aiming to improve astronomical observations and optical communication. The study compared Gaussian Processes (GPs) and a normalizing-flow-based model called FloTS against deterministic recurrent neural network (RNN) and long short-term memory (LSTM) models. Results indicate that deep learning models, particularly LSTM and FloTS, achieved superior accuracy in predicting atmospheric seeing up to two hours in advance. FloTS also demonstrated strong capabilities in probabilistic forecasting and uncertainty quantification, offering more informative predictions than GPs. AI

IMPACT Advances probabilistic forecasting and uncertainty quantification in atmospheric seeing prediction, benefiting astronomy and optical communications.

RANK_REASON Academic paper detailing machine learning model benchmarking for a specific scientific application. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Flotsam
• Gaussian Processes
• long short-term memory
• Mauna Kea
• recurrent neural network

• paper
• model release

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