{"id":"arxiv/2611.23979","url":"https://arxiv.org/abs/2611.23979","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Optimal Deep Nonparametric Regression with Uniformly Bounded Lipschitz Constants","domain":"stat.ML","source":"arxiv","authors":["Jiaxin Shi","Ziyuan Chen","Jian Huang","Hansheng Wang"],"reviews":0,"subject":"stat.ML","summary":"Stability under input perturbations is necessary for deep neural network estimators in many applications involving noisy covariates or estimated features. Weight norm constraints are widely used to control sensitivity to input perturbations by bounding the Lipschitz constant of neural networks. However, existing approximation results under fixed bounds on standard norm products are insufficient to establish optimal statistical rates. In this paper, we establish a sharp approximation theory for deep ReLU networks under a split spectral norm product constraint. This theoretical result shows…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-11-18","reactions":{}}