Optimized Architectures for Kolmogorov-Arnold Networks

ArXi:2512.12448v2 Announce Type: replace Efforts to improve Kolmogoro--Arnold networks (KANs) with architectural enhancements have been stymied by the complexity those enhancements bring, undermining the interpretability that makes KANs attractive in the first place. Here we study overprovisioned architectures combined with sparsification, deep supervision, and depth selection, to learn compact, interpretable KANs without sacrificing accuracy.