What is it about?

Here we show how to combine the mathematical theory of tipping points with artificial intelligence into a hybrid approach that can provide early warning signals of tipping points and also say something about the new state of the system that lies beyond the tipping point.

Featured Image

Why is it important?

Early warning signals of tipping points are vital to anticipate system collapse or other sudden shifts. However, existing generic early warning indicators designed to work across all systems do not provide information on the state that lies beyond the tipping point. Our results show how deep learning algorithms (artificial intelligence) can provide early warning signals of tipping points in real-world systems. The algorithm predicts certain qualitative aspects of the new state, and is also more sensitive and generates fewer false positives than generic indicators. We use theory about system behavior near tipping points so that the algorithm does not require data from the study system but instead learns from a universe of possible models.

Perspectives

This work provides a very different approach to predict tipping points than current methods. Some current methods extrapolate based on empirical data from previous tipping points events, which is good except that the nature of forcing might be different this time around. Or, some approaches may build mechanistic models to predict the future, although we cannot be certain we have included all of the relevant details in those models to predict the tipping point. In contrast, this work tells us that the clues to predicting the tipping point are already present in the noise the system experiences just before the transition. We only have to figure out how to interpret the noise.

Chris Bauch
University of Waterloo

Read the Original

This page is a summary of: Deep learning for early warning signals of tipping points, Proceedings of the National Academy of Sciences, September 2021, Proceedings of the National Academy of Sciences,
DOI: 10.1073/pnas.2106140118.
You can read the full text:

Read

Contributors

The following have contributed to this page