Forward search value iteration for POMDPs

Guy Shani; Ronen I. Brafman; Solomon E. Shimony

Forward search value iteration for POMDPs

Research output: Contribution to journal › Conference article › peer-review

82 Scopus citations

Abstract

Recent scaling up of POMDP solvers towards realistic applications is largely due to point-based methods which quickly converge to an approximate solution formedium-sized problems. Of this family HSVI, which uses trial-based asynchronous value iteration, can handle the largest domains. In this paper we suggest a new algorithm, FSVI, that uses the underlying MDP to traverse the belief space towards rewards, finding sequences of useful back-ups, and show how it scales up better than HSVI on larger benchmarks.

Original language	English
Pages (from-to)	2619-2624
Number of pages	6
Journal	IJCAI International Joint Conference on Artificial Intelligence
State	Published - 1 Dec 2007
Event	20th International Joint Conference on Artificial Intelligence, IJCAI 2007 - Hyderabad, India Duration: 6 Jan 2007 → 12 Jan 2007

ASJC Scopus subject areas

Artificial Intelligence

Cite this

@article{ede2b3dba65842daae85046daf7e9264,

title = "Forward search value iteration for POMDPs",

abstract = "Recent scaling up of POMDP solvers towards realistic applications is largely due to point-based methods which quickly converge to an approximate solution formedium-sized problems. Of this family HSVI, which uses trial-based asynchronous value iteration, can handle the largest domains. In this paper we suggest a new algorithm, FSVI, that uses the underlying MDP to traverse the belief space towards rewards, finding sequences of useful back-ups, and show how it scales up better than HSVI on larger benchmarks.",

author = "Guy Shani and Brafman, \{Ronen I.\} and Shimony, \{Solomon E.\}",

year = "2007",

month = dec,

day = "1",

language = "English",

pages = "2619--2624",

journal = "IJCAI International Joint Conference on Artificial Intelligence",

issn = "1045-0823",

publisher = "International Joint Conferences on Artificial Intelligence",

note = "20th International Joint Conference on Artificial Intelligence, IJCAI 2007 ; Conference date: 06-01-2007 Through 12-01-2007",

}

TY - JOUR

T1 - Forward search value iteration for POMDPs

AU - Shani, Guy

AU - Brafman, Ronen I.

AU - Shimony, Solomon E.

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Recent scaling up of POMDP solvers towards realistic applications is largely due to point-based methods which quickly converge to an approximate solution formedium-sized problems. Of this family HSVI, which uses trial-based asynchronous value iteration, can handle the largest domains. In this paper we suggest a new algorithm, FSVI, that uses the underlying MDP to traverse the belief space towards rewards, finding sequences of useful back-ups, and show how it scales up better than HSVI on larger benchmarks.

AB - Recent scaling up of POMDP solvers towards realistic applications is largely due to point-based methods which quickly converge to an approximate solution formedium-sized problems. Of this family HSVI, which uses trial-based asynchronous value iteration, can handle the largest domains. In this paper we suggest a new algorithm, FSVI, that uses the underlying MDP to traverse the belief space towards rewards, finding sequences of useful back-ups, and show how it scales up better than HSVI on larger benchmarks.

UR - https://www.scopus.com/pages/publications/84880906197

M3 - Conference article

AN - SCOPUS:84880906197

SN - 1045-0823

SP - 2619

EP - 2624

JO - IJCAI International Joint Conference on Artificial Intelligence

JF - IJCAI International Joint Conference on Artificial Intelligence

T2 - 20th International Joint Conference on Artificial Intelligence, IJCAI 2007

Y2 - 6 January 2007 through 12 January 2007

ER -

Forward search value iteration for POMDPs

Abstract

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this