SINDy-KANs combine Kolmogorov-Arnold Networks with Sparse Identification of Non-linear Dynamics to create parsimonious, interpretable models.
March 20, 2026
Original Paper
SINDy-KANs: Sparse identification of non-linear dynamics through Kolmogorov-Arnold networks
arXiv · 2603.18548
The Takeaway
While KANs were proposed for interpretability, they often lack sparsity; this work applies symbolic regression techniques at the activation level to discover actual sparse equations. This is a significant step toward making deep learning models truly transparent for scientific discovery.
From the abstract
Kolmogorov-Arnold networks (KANs) have arisen as a potential way to enhance the interpretability of machine learning. However, solutions learned by KANs are not necessarily interpretable, in the sense of being sparse or parsimonious. Sparse identification of nonlinear dynamics (SINDy) is a complementary approach that allows for learning sparse equations for dynamical systems from data; however, learned equations are limited by the library. In this work, we present SINDy-KANs, which simultaneousl