Optimal sensing via multi-armed bandit relaxations in mixed observability domains

Mikko Lauri; Risto Ritala

doi:10.1109/ICRA.2015.7139867

Optimal sensing via multi-armed bandit relaxations in mixed observability domains

Mikko Lauri
, Risto Ritala

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

6 Citations (Scopus)

44 Downloads (Pure)

Abstract

Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain.

Original language	English
Title of host publication	2015 IEEE International Conference on Robotics and Automation (ICRA), 26-30 May 2015, Seattle, WA
Pages	4807-4812
Number of pages	6
Volume	2015-June
DOIs	https://doi.org/10.1109/ICRA.2015.7139867
Publication status	Published - 29 Jun 2015
Publication type	A4 Article in conference proceedings
Event	IEEE International Conference on Robotics and Automation - Duration: 1 Jan 1900 → 1 Jan 2000

Conference

Conference	IEEE International Conference on Robotics and Automation
Period	1/01/00 → 1/01/00

Publication forum classification

Publication forum level 1

ASJC Scopus subject areas

Software
Artificial Intelligence
Control and Systems Engineering
Electrical and Electronic Engineering

Access to Document

10.1109/ICRA.2015.7139867

ICRA15_1071_FI
© 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 341 KB

http://urn.fi/URN:NBN:fi:tty-201605033926

Cite this

@inproceedings{995f61d0ea9a4bab9205889c1661f91f,

title = "Optimal sensing via multi-armed bandit relaxations in mixed observability domains",

abstract = "Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain.",

author = "Mikko Lauri and Risto Ritala",

year = "2015",

month = jun,

day = "29",

doi = "10.1109/ICRA.2015.7139867",

language = "English",

isbn = "978-1-4799-6923-4",

volume = "2015-June",

pages = "4807--4812",

booktitle = "2015 IEEE International Conference on Robotics and Automation (ICRA), 26-30 May 2015, Seattle, WA",

note = "IEEE International Conference on Robotics and Automation ; Conference date: 01-01-1900 Through 01-01-2000",

}

TY - GEN

T1 - Optimal sensing via multi-armed bandit relaxations in mixed observability domains

AU - Lauri, Mikko

AU - Ritala, Risto

PY - 2015/6/29

Y1 - 2015/6/29

N2 - Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain.

AB - Sequential decision making under uncertainty is studied in a mixed observability domain. The goal is to maximize the amount of information obtained on a partially observable stochastic process under constraints imposed by a fully observable internal state. An upper bound for the optimal value function is derived by relaxing constraints. We identify conditions under which the relaxed problem is a multi-armed bandit whose optimal policy is easily computable. The upper bound is applied to prune the search space in the original problem, and the effect on solution quality is assessed via simulation experiments. Empirical results show effective pruning of the search space in a target monitoring domain.

U2 - 10.1109/ICRA.2015.7139867

DO - 10.1109/ICRA.2015.7139867

M3 - Conference contribution

AN - SCOPUS:84938230088

SN - 978-1-4799-6923-4

VL - 2015-June

SP - 4807

EP - 4812

BT - 2015 IEEE International Conference on Robotics and Automation (ICRA), 26-30 May 2015, Seattle, WA

T2 - IEEE International Conference on Robotics and Automation

Y2 - 1 January 1900 through 1 January 2000

ER -

Optimal sensing via multi-armed bandit relaxations in mixed observability domains

Abstract

Conference

Publication forum classification

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this