KAN Layer based on MATLAB Deep Learning Toolbox

Kolmogorov-Arnold Networks (KAN) constructed based on MATLAB Deep Learning Toolbox.

Kolmogorov-Arnold Networks (KANs) are theoretically grounded alternatives to conventional Multilayer Perceptrons (MLPs). KANs are built upon the Kolmogorov-Arnold representation theorem, a profound mathematical result stating that any multivariate continuous function can be expressed through a finite composition of univariate continuous functions and binary additions. For more details about KANs, the interested readers can refer to the KAN: Kolmogorov-Arnold Networks.

The KAN Layer built here is based on the MATLAB Deep Learning Toolbox and exploit its power automatic differentiation capability provided by dlarray. It is worth noting that the current KAN Layer only receives feature input (e.g., feature vectors with "CB" format) and outputs same data format ("CB"). If you want to embed this KAN layer in deep learning pipeline, the input data format from the last layer must be "CB".