---
format: "aidr-story-markdown/v1"
id: "2b63c7c697dbbe1e89666b1ead5eed87015ae13599a77b9f68c63e8bd8d698f9"
canonical_url: "https://aidr.today/2b63c7c6?lang=en"
title: "Approximation Property of Dropout Neural Networks: Sobolev Rates and Confidence Bounds"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-05T04:00:00.000Z"
category: "Research"
topics: ["research"]
source_urls: ["https://arxiv.org/abs/2610.02253"]
summary: "The study analyzed approximation of the unit ball of $W^{n,\\infty}([0,1]^d)$ by ReLU networks whose edges are retained independently with probability $p$, constructing networks of constant depth and size $\\widetilde O_{n,d}(p^{-9}\\varepsilon^{-\\max\\{d/n,2\\}} \\log(1/\\delta))$ that guarantee approximation error bounds with probability at least $1-\\delta$."
---

# Approximation Property of Dropout Neural Networks: Sobolev Rates and Confidence Bounds

> [Open the canonical story](<https://aidr.today/2b63c7c6?lang=en>)

**Published:** 2026-10-05T04:00:00.000Z
**Category:** Research
**Topics:** research

## Summary

The study analyzed approximation of the unit ball of $W^\{n,\\infty\}\(\[0,1\]^d\)$ by ReLU networks whose edges are retained independently with probability $p$, constructing networks of constant depth and size $\\widetilde O\_\{n,d\}\(p^\{\-9\}\\varepsilon^\{\-\\max\\\{d/n,2\\\}\} \\log\(1/\\delta\)\)$ that guarantee approximation error bounds with probability at least $1\-\\delta$\.

## Sources

- [Story source](<https://arxiv.org/abs/2610.02253>)

