Skip to main navigation Skip to search Skip to main content

COMBINED NEURAL NETWORKS FOR TIME SERIES ANALYSIS

  • Tel Aviv University

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

65 Scopus citations

Abstract

We propose a method for improving the performance of any network designed to predict the next value of a time series. We advocate analyzing the deviations of the network's predictions from the data in the training set. This can be carried out by a secondary network trained on the time series of these residuals. The combined system of the two networks is viewed as the new predictor. We demonstrate the simplicity and success of this method, by applying it to the sunspots data. The small corrections of the secondary network can be regarded as resulting from a Taylor expansion of a complex network which includes the combined system. We find that the complex network is more difficult to train and performs worse than the two-step procedure of the combined system.

Original languageEnglish
Title of host publicationAdvances in Neural Information Processing Systems 6, NIPS 1993
EditorsJ. Cowan, G. Tesauro, J. Alspector
PublisherNeural information processing systems foundation
Pages224-231
Number of pages8
ISBN (Electronic)1558603220, 9781558603226
StatePublished - 1993
Event6th Advances in Neural Information Processing Systems, NIPS 1993 - Denver, United States
Duration: 29 Nov 19932 Dec 1993

Publication series

NameAdvances in Neural Information Processing Systems
Volume6
ISSN (Print)1049-5258

Conference

Conference6th Advances in Neural Information Processing Systems, NIPS 1993
Country/TerritoryUnited States
CityDenver
Period29/11/932/12/93

Fingerprint

Dive into the research topics of 'COMBINED NEURAL NETWORKS FOR TIME SERIES ANALYSIS'. Together they form a unique fingerprint.

Cite this