Sequential prediction of unbounded stationary time series

L. Györfi, György Ottucsák

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

A simple on-line procedure is considered for the prediction of a real-valued sequence. The algorithm is based on a combination of several simple predictors. If the sequence is a realization of an unbounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. An analog result is offered for the classification of binary processes.

Original languageEnglish
Pages (from-to)1866-1872
Number of pages7
JournalIEEE Transactions on Information Theory
Volume53
Issue number5
DOIs
Publication statusPublished - May 2007

Fingerprint

Random processes
time series
Time series

Keywords

  • On-line learning
  • Pattern recognition
  • Sequential prediction
  • Time series
  • Universal consistency

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Information Systems

Cite this

Sequential prediction of unbounded stationary time series. / Györfi, L.; Ottucsák, György.

In: IEEE Transactions on Information Theory, Vol. 53, No. 5, 05.2007, p. 1866-1872.

Research output: Contribution to journalArticle

@article{8871fa951c67409bb258183deb2eabfa,
title = "Sequential prediction of unbounded stationary time series",
abstract = "A simple on-line procedure is considered for the prediction of a real-valued sequence. The algorithm is based on a combination of several simple predictors. If the sequence is a realization of an unbounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. An analog result is offered for the classification of binary processes.",
keywords = "On-line learning, Pattern recognition, Sequential prediction, Time series, Universal consistency",
author = "L. Gy{\"o}rfi and Gy{\"o}rgy Ottucs{\'a}k",
year = "2007",
month = "5",
doi = "10.1109/TIT.2007.894660",
language = "English",
volume = "53",
pages = "1866--1872",
journal = "IEEE Transactions on Information Theory",
issn = "0018-9448",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "5",

}

TY - JOUR

T1 - Sequential prediction of unbounded stationary time series

AU - Györfi, L.

AU - Ottucsák, György

PY - 2007/5

Y1 - 2007/5

N2 - A simple on-line procedure is considered for the prediction of a real-valued sequence. The algorithm is based on a combination of several simple predictors. If the sequence is a realization of an unbounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. An analog result is offered for the classification of binary processes.

AB - A simple on-line procedure is considered for the prediction of a real-valued sequence. The algorithm is based on a combination of several simple predictors. If the sequence is a realization of an unbounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. An analog result is offered for the classification of binary processes.

KW - On-line learning

KW - Pattern recognition

KW - Sequential prediction

KW - Time series

KW - Universal consistency

UR - http://www.scopus.com/inward/record.url?scp=34248596750&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34248596750&partnerID=8YFLogxK

U2 - 10.1109/TIT.2007.894660

DO - 10.1109/TIT.2007.894660

M3 - Article

AN - SCOPUS:34248596750

VL - 53

SP - 1866

EP - 1872

JO - IEEE Transactions on Information Theory

JF - IEEE Transactions on Information Theory

SN - 0018-9448

IS - 5

ER -