What NARX networks can compute

Bill G. Horne, Hava T. Siegelmann, C. Lee Giles

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations

Abstract

We prove that a class of architectures called NARX neural networks, popular in control applications and other problems, are at least as powerful as fully connected recurrent neural networks. Recent results have shown that fully connected networks are Turing equivalent. Building on those results, we prove that NARX networks are also universal computation devices. NARX networks have a limited feedback which comes only from the output neuron rather than from hidden states. There is much interest in the amount and type of recurrence to be used in recurrent neural networks. Our results pose the question of what amount of feedback or recurrence is necessary for any network to be Turing equivalent and what restrictions on feedback limit computational power.

Original languageEnglish (US)
Title of host publicationSOFSEM 1995
Subtitle of host publicationTheory and Practice of Informatics - 22nd Seminar on Current Trends in Theory and Practice of Informatics, Proceedings
EditorsMiroslav Bartosek, Jan Staudek, Jiri Wiedermann
PublisherSpringer Verlag
Pages95-102
Number of pages8
ISBN (Print)3540606092, 9783540606093
DOIs
StatePublished - 1995
Event22nd International Seminar on Current Trends in Theory and Practice of Informatics, SOFSEM 1995 - Milovy, Czech Republic
Duration: Nov 23 1995Dec 1 1995

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1012
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other22nd International Seminar on Current Trends in Theory and Practice of Informatics, SOFSEM 1995
Country/TerritoryCzech Republic
CityMilovy
Period11/23/9512/1/95

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'What NARX networks can compute'. Together they form a unique fingerprint.

Cite this