Field Theoretical Analysis of On-Line Learning of Probability Distributions

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since a renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model.

Original languageEnglish
Pages (from-to)3554-3557
Number of pages4
JournalPhysical Review Letters
Volume83
Issue number17
Publication statusPublished - Oct 25 1999
Externally publishedYes

Fingerprint

learning
degrees of freedom

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Field Theoretical Analysis of On-Line Learning of Probability Distributions. / Aida, Toshiaki.

In: Physical Review Letters, Vol. 83, No. 17, 25.10.1999, p. 3554-3557.

Research output: Contribution to journalArticle

@article{02af7b4b38564d15b7c8a7c23603824f,
title = "Field Theoretical Analysis of On-Line Learning of Probability Distributions",
abstract = "On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since a renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model.",
author = "Toshiaki Aida",
year = "1999",
month = "10",
day = "25",
language = "English",
volume = "83",
pages = "3554--3557",
journal = "Physical Review Letters",
issn = "0031-9007",
publisher = "American Physical Society",
number = "17",

}

TY - JOUR

T1 - Field Theoretical Analysis of On-Line Learning of Probability Distributions

AU - Aida, Toshiaki

PY - 1999/10/25

Y1 - 1999/10/25

N2 - On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since a renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model.

AB - On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since a renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model.

UR - http://www.scopus.com/inward/record.url?scp=0000430886&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0000430886&partnerID=8YFLogxK

M3 - Article

VL - 83

SP - 3554

EP - 3557

JO - Physical Review Letters

JF - Physical Review Letters

SN - 0031-9007

IS - 17

ER -