KernelOnet: An Interpretable Neural Operator Based on Kernel Functions

License: See LICENSE file · Updated 2026-09-28
My rating

This paper proposes KernelOnet, an interpretable neural operator framework that explicitly incorporates kernel functions into the operator architecture. Its structure is designed to match the kernel-expansion form used in boundary-type kernel-expansion methods, while replacing the implicitly learned basis functions of traditional deep-network operators such as DeepONet with explicit kernels.

Tags: