Export

Link to this page

Please use this URL for citations.

Search Other Services

Statistics

  • Access to this page：962 times

＜journal article＞
NUMERICAL ENCLOSURE FOR THE OPTIMAL THRESHOLD PROBABILITY IN DISCOUNTED MARKOV DECISION PROCESSES

Creator
Creator Name
Toyonaga, Kenji
豊永, 憲治
Affiliation
Affiliation Name
Graduate School of Mathematics, Kyushu University
九州大学大学院数理学府
Author PID
K000694
Creator Name
Nakao, T. Mitsuhiro
中尾, 充宏
Affiliation
Affiliation Name
Graduate School of Mathematics, Kyushu University
九州大学大学院数理学府
Language
English
Publisher
Research Association of Statistical Sciences
統計科学研究会
Date
2000-06
Source Title
Bulletin of informatics and cybernetics
Vol
32
Issue
1
First Page
81
Last Page
90
Publication Type
Version of Record
Access Rights
open access
Crossref DOI
https://doi.org/10.5109/13494
Related DOI
Bulletin of informatics and cybernetics || 32(1) || p81-90
http://bic.math.kyushu-u.ac.jp/
Related URI
Bulletin of informatics and cybernetics || 32(1) || p81-90
http://bic.math.kyushu-u.ac.jp/
Relation
Bulletin of informatics and cybernetics || 32(1) || p81-90
http://bic.math.kyushu-u.ac.jp/
Abstract There are various procedures to compute the optimal threshold probability in discounted Markov decision processes. In the actual numerical computation of an approximate optimal solution, the estimatio...n of the discrepancy between the approximate solution and the exact solution is important. White(1993 b) derived such an error estimation for the value iteration method, however, this estimation is not actually in the computable form. In this paper, we present a numerical enclosure method to compute the optimal threshold probability, that guarantees a rigorous a posteriori error bound.show more

Hide fulltext details.

pdf p081 pdf 192 KB 340  

Details

PISSN
0286-522X
EISSN
2435-743X
NCID
AA10634475
Record ID
13494
Peer-Reviewed
Refereed
Type
学術雑誌論文
Created Date 2009.04.22
Modified Date 2020.10.22

People who viewed this item also viewed