{"id":"arxiv/2606.23575","url":"https://arxiv.org/abs/2606.23575","tags":["cs.LG","cs.AI","math.NA","stat.ML"],"score":0,"title":"Solve for the Hyperparameter, Skip the Search: Kolmogorov-Optimal Scaling Laws for Spline Regression","domain":"cs.LG","source":"arxiv","authors":["Yong Yi Bay","Kathleen A. Yearick"],"reviews":0,"subject":"cs.LG","summary":"Hyperparameter tuning almost always means search: fit the model at every value on a grid, score each by cross-validation, and keep the winner. For spline regression that search is unnecessary. The optimal resolution can be solved for in closed form, to the accuracy an exhaustive search reaches, at a fraction of the compute. Three ingredients make this possible: classical approximation theory pins the squared bias to a known power of the resolution G, exactly the Kolmogorov n-width of the smoothness class; the basis dimension is an explicit polynomial in G; and leave-one-out error follows…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-06-22","reactions":{}}