Some pre-trained artificial neural network have shown to be able to predict transitions (and their underlying normal form). I suppose it works well if data corresponds to training set (relatively large for the below-mentioned reference), worse if it is not the case.
Reference
Bury et al. (2021), Deep learning for early warning signals of tipping points.
Codebase
Rely on some deep-learning library in julia (Flux.jl standard but canbe discussed)
Indicator summary
Some pre-trained artificial neural network have shown to be able to predict transitions (and their underlying normal form). I suppose it works well if data corresponds to training set (relatively large for the below-mentioned reference), worse if it is not the case.
Reference
Bury et al. (2021), Deep learning for early warning signals of tipping points.
Codebase
Rely on some deep-learning library in julia (Flux.jl standard but canbe discussed)
Implementation plan
Find and incorporate pre-trained network.