English

State-Space Kolmogorov Arnold Networks for Interpretable Nonlinear System Identification

Machine Learning 2025-06-24 v1 Systems and Control Systems and Control

Abstract

While accurate, black-box system identification models lack interpretability of the underlying system dynamics. This paper proposes State-Space Kolmogorov-Arnold Networks (SS-KAN) to address this challenge by integrating Kolmogorov-Arnold Networks within a state-space framework. The proposed model is validated on two benchmark systems: the Silverbox and the Wiener-Hammerstein benchmarks. Results show that SS-KAN provides enhanced interpretability due to sparsity-promoting regularization and the direct visualization of its learned univariate functions, which reveal system nonlinearities at the cost of accuracy when compared to state-of-the-art black-box models, highlighting SS-KAN as a promising approach for interpretable nonlinear system identification, balancing accuracy and interpretability of nonlinear system dynamics.

Keywords

Cite

@article{arxiv.2506.16392,
  title  = {State-Space Kolmogorov Arnold Networks for Interpretable Nonlinear System Identification},
  author = {Gonçalo Granjal Cruz and Balazs Renczes and Mark C Runacres and Jan Decuyper},
  journal= {arXiv preprint arXiv:2506.16392},
  year   = {2025}
}

Comments

Accepted for IEEE Control Systems Letters

R2 v1 2026-07-01T03:25:19.286Z