SINDy
A sparse symbolic regression method proposed by Brunton et al. in 2016. It performs sparse optimization over a predefined library of candidate functions to directly output human-readable differential equations. The method offers strong interpretability and is widely used for nonlinear dynamical system identification.
Timeline (last 90 days)
Symbolic regression is a class of methods that automatically discover mathematical expressions from data, with typical tools including genetic programming and sparse regression (such as SINDy).
SINDy (Sparse Identification of Nonlinear Dynamics) was proposed by Brunton et al. in 2016.
The core idea of SINDy is to find, through sparse optimization within a pre-specified library of candidate functions, the differential equation terms that best explain the observational data, and to directly output human-readable equations.
SINDy has extremely strong interpretability but relies on prior design of the function library.
Both SINDy and Neural ODE require training on specific systems and do not possess native zero-shot transfer capability.
All Facts (5)
Symbolic regression is a class of methods that automatically discover mathematical expressions from data, with typical tools including genetic programming and sparse regression (such as SINDy).
50%UnverifiedSINDy (Sparse Identification of Nonlinear Dynamics) was proposed by Brunton et al. in 2016.
50%UnverifiedThe core idea of SINDy is to find, through sparse optimization within a pre-specified library of candidate functions, the differential equation terms that best explain the observational data, and to directly output human-readable equations.
50%UnverifiedSINDy has extremely strong interpretability but relies on prior design of the function library.
50%UnverifiedBoth SINDy and Neural ODE require training on specific systems and do not possess native zero-shot transfer capability.
50%