Predicting critical transitions with machine learning trained on surrogates of historical data

Critical transitions can occur in many natural and man-made systems. Generic early warning signals motivated by dynamical systems theory have had mixed success on real noisy data. More recent studies found that deep learning classifiers trained on synthetic data could improve performance. However, to the best of our knowledge, neither of these methods take advantage of historical, system-specific data. Here, we introduce an approach that trains machine learning classifiers on surrogate data of past transitions. The approach provides early warning signals in empirical and experimental data with higher sensitivity and specificity than two widely used generic early warning signals—variance and lag-1 autocorrelation. Since the approach is trained on surrogates of historical data, it is not bound by the restricting assumption of a local bifurcation like previous methods. This system-specific approach can contribute to improved early warning signals to help humans better prepare for or avoid undesirable critical transitions. Critical transitions often lead to catastrophic regime shifts, so reliable early warning signals are crucial to prevent irreversible damage. Here, the authors present a system-specific framework for early warning of tipping points in complex systems via training machine learning models on surrogate data extracted from limited availability historical records.