Reproduce “Tractable Inference with Kolmogorov-Arnold Networks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Juan Diego Toscano, Sascha Ranftl, Zhiwei Gao, Li-Lian Wang, George Em Karniadakis
TOUCKANs compute the mean and variance of a trained scientific surrogate through deterministic recursions. Polynomial functions of sums of one-variable branches expand into products whose integrals separate under independent input coordinates. On the Oakley–O’Hagan-type benchmark at D = 25, mean and variance relative errors were 4.71 × 10⁻⁷ and 1.02 × 10⁻⁶. The method assumes factorizing inputs; accuracy also depends on surrogate fit and quadrature.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.