Bayesian Approach to the Optimization of Adaptive Systems

Ta Tung Lin, Sik-Sang Yau

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

This paper describes how an adaptive system can adapt itself to optimize its performance under the influence of uncertain environment. At each stage of adaptation, the uncertain environment, which is represented by a random vector with an unknown statistical property, is estimated by Bayesian approach from its past outcomes up to the latest one. This approach is investigated in general so that the probability distribution of the future outcomes of the random vector is not restricted to any particular one. For most of the adaptive systems, these probability distributions are assumed to be the same. However, in the case of signal adaptation, it is shown that the results as well as the execution of the optimization technique are alike whether or not the probability distributions of the forthcoming outcomes of the random vector are the same.

Original languageEnglish (US)
Pages (from-to)77-85
Number of pages9
JournalIEEE Transactions on Systems Science and Cybernetics
Volume3
Issue number2
DOIs
StatePublished - 1967
Externally publishedYes

Fingerprint

Adaptive systems
Probability distributions

ASJC Scopus subject areas

  • Computer Science Applications
  • Human-Computer Interaction
  • Software
  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Cite this

Bayesian Approach to the Optimization of Adaptive Systems. / Lin, Ta Tung; Yau, Sik-Sang.

In: IEEE Transactions on Systems Science and Cybernetics, Vol. 3, No. 2, 1967, p. 77-85.

Research output: Contribution to journalArticle

@article{24a1e0433ac44e2fadd046e8d83587ff,
title = "Bayesian Approach to the Optimization of Adaptive Systems",
abstract = "This paper describes how an adaptive system can adapt itself to optimize its performance under the influence of uncertain environment. At each stage of adaptation, the uncertain environment, which is represented by a random vector with an unknown statistical property, is estimated by Bayesian approach from its past outcomes up to the latest one. This approach is investigated in general so that the probability distribution of the future outcomes of the random vector is not restricted to any particular one. For most of the adaptive systems, these probability distributions are assumed to be the same. However, in the case of signal adaptation, it is shown that the results as well as the execution of the optimization technique are alike whether or not the probability distributions of the forthcoming outcomes of the random vector are the same.",
author = "Lin, {Ta Tung} and Sik-Sang Yau",
year = "1967",
doi = "10.1109/TSSC.1967.300086",
language = "English (US)",
volume = "3",
pages = "77--85",
journal = "IEEE Transactions on Systems Science and Cybernetics",
issn = "0536-1567",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "2",

}

TY - JOUR

T1 - Bayesian Approach to the Optimization of Adaptive Systems

AU - Lin, Ta Tung

AU - Yau, Sik-Sang

PY - 1967

Y1 - 1967

N2 - This paper describes how an adaptive system can adapt itself to optimize its performance under the influence of uncertain environment. At each stage of adaptation, the uncertain environment, which is represented by a random vector with an unknown statistical property, is estimated by Bayesian approach from its past outcomes up to the latest one. This approach is investigated in general so that the probability distribution of the future outcomes of the random vector is not restricted to any particular one. For most of the adaptive systems, these probability distributions are assumed to be the same. However, in the case of signal adaptation, it is shown that the results as well as the execution of the optimization technique are alike whether or not the probability distributions of the forthcoming outcomes of the random vector are the same.

AB - This paper describes how an adaptive system can adapt itself to optimize its performance under the influence of uncertain environment. At each stage of adaptation, the uncertain environment, which is represented by a random vector with an unknown statistical property, is estimated by Bayesian approach from its past outcomes up to the latest one. This approach is investigated in general so that the probability distribution of the future outcomes of the random vector is not restricted to any particular one. For most of the adaptive systems, these probability distributions are assumed to be the same. However, in the case of signal adaptation, it is shown that the results as well as the execution of the optimization technique are alike whether or not the probability distributions of the forthcoming outcomes of the random vector are the same.

UR - http://www.scopus.com/inward/record.url?scp=84913082163&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84913082163&partnerID=8YFLogxK

U2 - 10.1109/TSSC.1967.300086

DO - 10.1109/TSSC.1967.300086

M3 - Article

VL - 3

SP - 77

EP - 85

JO - IEEE Transactions on Systems Science and Cybernetics

JF - IEEE Transactions on Systems Science and Cybernetics

SN - 0536-1567

IS - 2

ER -