Market-based reinforcement learning in partially observable worlds

Ivo Kwee; Marcus Hutter; Jürgen Schmidhuber

doi:10.1007/3-540-44668-0_120

Market-based reinforcement learning in partially observable worlds

Ivo Kwee, Marcus Hutter, Jürgen Schmidhuber

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

11 Scopus citations

Abstract

Unlike traditional reinforcement learning (RL), marketbased RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.

Original language	English (US)
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher	Springer [email protected]
Pages	865-873
Number of pages	9
ISBN (Print)	3540424865
DOIs	https://doi.org/10.1007/3-540-44668-0_120
State	Published - Jan 1 2001
Externally published	Yes

Access to Document

10.1007/3-540-44668-0_120

Cite this

@inproceedings{e66d2bda28354c32aab25958867a4c78,

title = "Market-based reinforcement learning in partially observable worlds",

abstract = "Unlike traditional reinforcement learning (RL), marketbased RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.",

author = "Ivo Kwee and Marcus Hutter and J{\"u}rgen Schmidhuber",

note = "Generated from Scopus record by KAUST IRTS on 2022-09-14",

year = "2001",

month = jan,

day = "1",

doi = "10.1007/3-540-44668-0_120",

language = "English (US)",

isbn = "3540424865",

pages = "865--873",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer [email protected]",

}

Market-based reinforcement learning in partially observable worlds. / Kwee, Ivo; Hutter, Marcus; Schmidhuber, Jürgen.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Springer [email protected], 2001. p. 865-873.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Market-based reinforcement learning in partially observable worlds

AU - Kwee, Ivo

AU - Hutter, Marcus

AU - Schmidhuber, Jürgen

N1 - Generated from Scopus record by KAUST IRTS on 2022-09-14

PY - 2001/1/1

Y1 - 2001/1/1

N2 - Unlike traditional reinforcement learning (RL), marketbased RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.

AB - Unlike traditional reinforcement learning (RL), marketbased RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.

UR - http://link.springer.com/10.1007/3-540-44668-0_120

UR - http://www.scopus.com/inward/record.url?scp=84865410615&partnerID=8YFLogxK

U2 - 10.1007/3-540-44668-0_120

DO - 10.1007/3-540-44668-0_120

M3 - Conference contribution

SN - 3540424865

SP - 865

EP - 873

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer [email protected]

ER -

Market-based reinforcement learning in partially observable worlds

Abstract

Access to Document

Other files and links

Fingerprint

Cite this