Machine learning methods trained on simple models can predict critical transitions in complex natural systems
Forecasting sudden changes in complex systems is a critical but challenging task, with previously developed methods varying widely in their reliability. Here we develop a novel detection method, using simple theoretical models to train a deep neural network to detect critical transitions—the Early W...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
The Royal Society
2022-02-01
|
Series: | Royal Society Open Science |
Subjects: | |
Online Access: | https://royalsocietypublishing.org/doi/10.1098/rsos.211475 |