04:00
2026-10-05
arxiv.org
machine-learning
Approximation Property of Dropout Neural Networks: Sobolev Rates and Confidence Bounds
A new arXiv paper (2610.02253v1) establishes approximation rates for dropout ReLU networks, showing that the unit ball of W^{n,∞}([0,1]^d) can be approximated uniformly by constant-depth networks of s…