Can random linear networks store multiple long input streams?

Adam S. Charles, Dong Yin, Christopher J. Rozell

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

Abstract

The short term memory of randomly connected networks has been recently studied in order to better understand the computational and predictive power of such networks. In particular, random, linear, orthogonal networks have been explored extensively in the context a single input stream driving the network. The most recent results state that a stream of length N can be recovered from a network of size 0(S log6 (N)) assuming that the input is S-sparse in some basis. Little work, however, addresses more complex networks where multiple input streams feed into the same network. In this paper we extend the results for recovering sparse input streams the multiple input streams feeding into the same network. We find that we can recover L input streams of length N with a network that has O (S log5 (LN)) nodes.

Original languageEnglish (US)
Title of host publication2014 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2014
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages379-383
Number of pages5
ISBN (Electronic)9781479970889
DOIs
StatePublished - Feb 5 2014
Externally publishedYes
Event2014 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2014 - Atlanta, United States
Duration: Dec 3 2014Dec 5 2014

Publication series

Name2014 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2014

Conference

Conference2014 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2014
Country/TerritoryUnited States
CityAtlanta
Period12/3/1412/5/14

Keywords

  • Linear neural network
  • Restricted isometry constant
  • Short-term memory
  • Sparse signals

ASJC Scopus subject areas

  • Signal Processing
  • Information Systems

Fingerprint

Dive into the research topics of 'Can random linear networks store multiple long input streams?'. Together they form a unique fingerprint.

Cite this