State-Space Kolmogorov Arnold Networks for Interpretable Nonlinear System Identification
This work addresses the need for interpretable models in nonlinear system identification, particularly for domains requiring insights into system dynamics, but it is incremental as it builds on existing Kolmogorov-Arnold Networks within a state-space context.
The paper tackled the problem of interpretability in black-box system identification models by proposing State-Space Kolmogorov-Arnold Networks (SS-KAN), which integrate Kolmogorov-Arnold Networks into a state-space framework; results on benchmark systems showed enhanced interpretability through sparsity and visualization of learned functions, though with some accuracy trade-offs compared to state-of-the-art models.
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.