Universal Approximation in Dropout Neural Networks

Research output: Contribution to journalArticleAcademicpeer-review

1 Downloads (Pure)

Abstract

We prove two universal approximation theorems for a range of dropout neural networks. These are feed-forward neural networks in which each edge is given a random {0, 1}-valued filter, that have two modes of operation: in the first each edge output is multiplied by its random filter, resulting in a random output, while in the second each edge output is multiplied by the expectation of its filter, leading to a deterministic output. It is common to use the random mode during training and the deterministic mode during testing and prediction. Both theorems are of the following form: Given a function to approximate and a threshold ε > 0, there exists a dropout network that is ε-close in probability and in Lq. The first theorem applies to dropout networks in the random mode. It assumes little on the activation function, applies to a wide class of networks, and can even be applied to approximation schemes other than neural networks. The core is an algebraic property that shows that deterministic networks can be exactly matched in expectation by random networks. The second theorem makes stronger assumptions and gives a stronger result. Given a function to approximate, it provides existence of a network that approximates in both modes simultaneously. Proof components are a recursive replacement of edges by independent copies, and a special first-layer replacement that couples the resulting larger network to the input. The functions to be approximated are assumed to be elements of general normed spaces, and the approximations are measured in the corresponding norms. The networks are constructed explicitly. Because of the different methods of proof, the two results give independent insight into the approximation properties of random dropout networks. With this, we establish that dropout neural networks broadly satisfy a universal-approximation property.

Original languageEnglish
Pages (from-to)1-46
JournalJournal of Machine Learning Research
Volume23
Issue number19
Publication statusPublished - 1 Feb 2022

Bibliographical note

Funding Information:
J.W. Portegies was supported by the Electronic Component Systems for European Leadership Joint Undertaking under grant agreement No 737459 (project Productive 4.0). This Joint Undertaking receives support from the European Union Horizon 2020 research and innovation program and Germany, Austria, France, Czech Republic, Netherlands, Belgium, Spain, Greece, Sweden, Italy, Ireland, Poland, Hungary, Portugal, Denmark, Finland, Luxembourg, Norway, Turkey.

Publisher Copyright:
© 2022 Oxana A. Manita, Mark A. Peletier, Jacobus W. Portegies, Jaron Sanders and Albert Senen-Cerda..

Keywords

  • Approximation
  • Dropout
  • Neural networks
  • Random neural network

Fingerprint

Dive into the research topics of 'Universal Approximation in Dropout Neural Networks'. Together they form a unique fingerprint.

Cite this