DP-KAN: Differentially Private Kolmogorov-Arnold Networks

We study the Kolmogorov-Arnold Network (KAN), recently proposed as an alternative to the classical Multilayer Perceptron (MLP), in the application for differentially private model training. Using the DP-SGD algorithm, we demonstrate that KAN can be made private in a straightforward manner and evaluated its performance across several datasets. Our results indicate that the accuracy of KAN is not only comparable with MLP but also experiences similar deterioration due to privacy constraints, making it suitable for differentially private model training. The Kolmogorov-Arnold Network (KAN) (Ziming et al., 2024) has recently emerged as a new approach to symbolic regression and general prediction problems. This new architecture already saw remarkable attention in very recent papers (Shukla et al., 2024; Genet & Inzirillo, 2024; Zavareh & Chen, 2024; Vaca-Rubio et al., 2024; Xu et al., 2024).