Symbolic regression

Can we, or machines, identify the mathematical law underlying the dynamics that we see? Schmidt2009distilling demonstrated that this is possible.

Fajardo Fontiveros2023fundamental argued that there is a transition between low-noise, learnable phase and high-noise phase where the closed-form model cannot be learned.

How about using Artificial neural network (Deep learning)? Some approaches exist (e.g., Kubalík2023neural).

Fajardo-Fontiveros2023fundamental studies fundamental limits.

See also SINDy