rKAN: Rational Kolmogorov-Arnold Networks
This work addresses implementation challenges in KANs for researchers in deep learning, offering an incremental improvement over existing basis functions.
The paper tackled the complexity of Kolmogorov-Arnold networks (KANs) by introducing rational functions as a novel basis function, proposing rKAN with Pade approximation and rational Jacobi functions, and demonstrated its effectiveness in deep learning and physics-informed tasks.
The development of Kolmogorov-Arnold networks (KANs) marks a significant shift from traditional multi-layer perceptrons in deep learning. Initially, KANs employed B-spline curves as their primary basis function, but their inherent complexity posed implementation challenges. Consequently, researchers have explored alternative basis functions such as Wavelets, Polynomials, and Fractional functions. In this research, we explore the use of rational functions as a novel basis function for KANs. We propose two different approaches based on Pade approximation and rational Jacobi functions as trainable basis functions, establishing the rational KAN (rKAN). We then evaluate rKAN's performance in various deep learning and physics-informed tasks to demonstrate its practicality and effectiveness in function approximation.