Predicting discrete-time bifurcations with deep learning
This work addresses the need for early warning signals in systems prone to discrete-time bifurcations, offering a novel tool for monitoring critical transitions in fields such as ecology and economics, though it is incremental as it extends existing deep learning methods from continuous-time to discrete-time bifurcations.
The authors tackled the problem of predicting discrete-time bifurcations, which are critical transitions in systems like physiology and economics, by training a deep learning classifier that outperforms common early warning signals under various noise conditions and accurately identifies specific bifurcation types.
Many natural and man-made systems are prone to critical transitions -- abrupt and potentially devastating changes in dynamics. Deep learning classifiers can provide an early warning signal (EWS) for critical transitions by learning generic features of bifurcations (dynamical instabilities) from large simulated training data sets. So far, classifiers have only been trained to predict continuous-time bifurcations, ignoring rich dynamics unique to discrete-time bifurcations. Here, we train a deep learning classifier to provide an EWS for the five local discrete-time bifurcations of codimension-1. We test the classifier on simulation data from discrete-time models used in physiology, economics and ecology, as well as experimental data of spontaneously beating chick-heart aggregates that undergo a period-doubling bifurcation. The classifier outperforms commonly used EWS under a wide range of noise intensities and rates of approach to the bifurcation. It also predicts the correct bifurcation in most cases, with particularly high accuracy for the period-doubling, Neimark-Sacker and fold bifurcations. Deep learning as a tool for bifurcation prediction is still in its nascence and has the potential to transform the way we monitor systems for critical transitions.