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 |
_version_ | 1797837659461124096 |
---|---|
author | Smita Deb Sahil Sidheekh Christopher F. Clements Narayanan C. Krishnan Partha S. Dutta |
author_facet | Smita Deb Sahil Sidheekh Christopher F. Clements Narayanan C. Krishnan Partha S. Dutta |
author_sort | Smita Deb |
collection | DOAJ |
description | 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 Warning Signal Network (EWSNet). We then demonstrate that this network, trained on simulated data, can reliably predict observed real-world transitions in systems ranging from rapid climatic change to the collapse of ecological populations. Importantly, our model appears to capture latent properties in time series missed by previous warning signals approaches, allowing us to not only detect if a transition is approaching, but critically whether the collapse will be catastrophic or non-catastrophic. These novel properties mean EWSNet has the potential to serve as an indicator of transitions across a broad spectrum of complex systems, without requiring information on the structure of the system being monitored. Our work highlights the practicality of deep learning for addressing further questions pertaining to ecosystem collapse and has much broader management implications. |
first_indexed | 2024-04-09T15:28:16Z |
format | Article |
id | doaj.art-b31b79479af841afa7b43bf5cc153e92 |
institution | Directory Open Access Journal |
issn | 2054-5703 |
language | English |
last_indexed | 2024-04-09T15:28:16Z |
publishDate | 2022-02-01 |
publisher | The Royal Society |
record_format | Article |
series | Royal Society Open Science |
spelling | doaj.art-b31b79479af841afa7b43bf5cc153e922023-04-28T11:04:41ZengThe Royal SocietyRoyal Society Open Science2054-57032022-02-019210.1098/rsos.211475Machine learning methods trained on simple models can predict critical transitions in complex natural systemsSmita Deb0Sahil Sidheekh1Christopher F. Clements2Narayanan C. Krishnan3Partha S. Dutta4Department of Mathematics, Indian Institute of Technology Ropar, Rupnagar, Punjab 140001, IndiaDepartment of Computer Science and Engineering, Indian Institute of Technology Ropar, Rupnagar, Punjab 140001, IndiaSchool of Biological Sciences, University of Bristol, Bristol BS8 1TQ, UKDepartment of Computer Science and Engineering, Indian Institute of Technology Ropar, Rupnagar, Punjab 140001, IndiaDepartment of Mathematics, Indian Institute of Technology Ropar, Rupnagar, Punjab 140001, IndiaForecasting 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 Warning Signal Network (EWSNet). We then demonstrate that this network, trained on simulated data, can reliably predict observed real-world transitions in systems ranging from rapid climatic change to the collapse of ecological populations. Importantly, our model appears to capture latent properties in time series missed by previous warning signals approaches, allowing us to not only detect if a transition is approaching, but critically whether the collapse will be catastrophic or non-catastrophic. These novel properties mean EWSNet has the potential to serve as an indicator of transitions across a broad spectrum of complex systems, without requiring information on the structure of the system being monitored. Our work highlights the practicality of deep learning for addressing further questions pertaining to ecosystem collapse and has much broader management implications.https://royalsocietypublishing.org/doi/10.1098/rsos.211475catastrophic transitionsnon-catastrophic transitionstipping pointsearly warning indicatorsclassificationdeep learning |
spellingShingle | Smita Deb Sahil Sidheekh Christopher F. Clements Narayanan C. Krishnan Partha S. Dutta Machine learning methods trained on simple models can predict critical transitions in complex natural systems Royal Society Open Science catastrophic transitions non-catastrophic transitions tipping points early warning indicators classification deep learning |
title | Machine learning methods trained on simple models can predict critical transitions in complex natural systems |
title_full | Machine learning methods trained on simple models can predict critical transitions in complex natural systems |
title_fullStr | Machine learning methods trained on simple models can predict critical transitions in complex natural systems |
title_full_unstemmed | Machine learning methods trained on simple models can predict critical transitions in complex natural systems |
title_short | Machine learning methods trained on simple models can predict critical transitions in complex natural systems |
title_sort | machine learning methods trained on simple models can predict critical transitions in complex natural systems |
topic | catastrophic transitions non-catastrophic transitions tipping points early warning indicators classification deep learning |
url | https://royalsocietypublishing.org/doi/10.1098/rsos.211475 |
work_keys_str_mv | AT smitadeb machinelearningmethodstrainedonsimplemodelscanpredictcriticaltransitionsincomplexnaturalsystems AT sahilsidheekh machinelearningmethodstrainedonsimplemodelscanpredictcriticaltransitionsincomplexnaturalsystems AT christopherfclements machinelearningmethodstrainedonsimplemodelscanpredictcriticaltransitionsincomplexnaturalsystems AT narayananckrishnan machinelearningmethodstrainedonsimplemodelscanpredictcriticaltransitionsincomplexnaturalsystems AT parthasdutta machinelearningmethodstrainedonsimplemodelscanpredictcriticaltransitionsincomplexnaturalsystems |