{"id":"arxiv/2611.03653","url":"https://arxiv.org/abs/2611.03653","tags":["cs.LG","math.PR","stat.ML"],"score":0,"title":"Four Extensions of the Universal Law of Robustness: Log-Concave Covariates, Heterogeneous Mixtures, Vector-Valued Predictors, and Markovian Data","domain":"cs.LG","source":"arxiv","authors":["Tobias J. Peter"],"reviews":0,"subject":"cs.LG","summary":"Bubeck and Sellke's universal law of robustness says that fitting $n$ noisy samples below the noise level with a class of $p$ real parameters forces a Lipschitz constant of order $\\sqrt{nd/p}$, when the covariate distribution on $\\mathbb{R}^d$ is isoperimetric: smooth interpolation costs a factor $d$ of overparametrization. Its hypotheses are Gaussian-type concentration, a homogeneous covariate law, scalar outputs, and independent samples; we relax each one. Exponential concentration suffices, at the unchanged rate and absolute constant and at the price of a capacity burn-in, some form of…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-08-26","reactions":{}}