{"id":"arxiv/2605.18598","url":"https://arxiv.org/abs/2605.18598","tags":["cs.LG","cond-mat.stat-mech","math.FA","math.PR","math.ST"],"score":0,"title":"Pointwise Generalization in Deep Neural Networks","domain":"cs.LG","source":"arxiv","authors":["Shaojie Li","Yunbei Xu"],"reviews":0,"subject":"cs.LG","summary":"We address the fundamental question of why deep neural networks generalize by establishing a pointwise generalization theory for fully connected networks. This framework resolves long-standing barriers to characterizing the rich nonlinear feature-learning regime and builds a new statistical foundation for representation learning. For each trained model, we characterize the hypothesis via a pointwise Riemannian Dimension, derived from the eigenvalues of the learned feature representations across layers. This establishes a principled framework for deriving hypothesis-dependent,…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-05-18","reactions":{}}