Relational Representations in Reinforcement Learning: Review and Open Problems

M. van Otterlo

Relational Representations in Reinforcement Learning: Review and Open Problems

M. van Otterlo

Research output: Contribution to conference › Paper › Academic › peer-review

196 Downloads (Pure)

Abstract

This paper is about representation in RL.We discuss some of the concepts in representation and generalization in reinforcement learning and argue for higher-order representations, instead of the commonly used propositional representations. The paper contains a small review of current reinforcement learning systems using higher-order representations, followed by a brief discussion. The paper ends with research directions and open problems.

Original language	English
Number of pages	8
Publication status	Published - 9 Jul 2002
Event	ICML-2002 Workshop on Development of Representations - University of New South Wales, Sydney, Australia Duration: 9 Jul 2002 → 9 Jul 2002 http://www.demo.cs.brandeis.edu/icml02ws/

Workshop

Workshop	ICML-2002 Workshop on Development of Representations
Country/Territory	Australia
City	Sydney
Period	9/07/02 → 9/07/02
Internet address	http://www.demo.cs.brandeis.edu/icml02ws/

Keywords

EWI-14945
Review
relational representations
HMI-IA: Intelligent Agents
IR-65334
Generalization
Reinforcement learning

Access to Document

VanOtterlo2002relationalFinal published version, 138 KB

http://demo.cs.brandeis.edu/icml02ws/otterlo.pdf

Cite this

@conference{add2381ce2214a98bf0587a3e4f01722,

title = "Relational Representations in Reinforcement Learning: Review and Open Problems",

abstract = "This paper is about representation in RL.We discuss some of the concepts in representation and generalization in reinforcement learning and argue for higher-order representations, instead of the commonly used propositional representations. The paper contains a small review of current reinforcement learning systems using higher-order representations, followed by a brief discussion. The paper ends with research directions and open problems.",

keywords = "EWI-14945, Review, relational representations, HMI-IA: Intelligent Agents, IR-65334, Generalization, Reinforcement learning",

author = "{van Otterlo}, M.",

year = "2002",

month = jul,

day = "9",

language = "English",

note = "ICML-2002 Workshop on Development of Representations ; Conference date: 09-07-2002 Through 09-07-2002",

url = "http://www.demo.cs.brandeis.edu/icml02ws/",

}

TY - CONF

T1 - Relational Representations in Reinforcement Learning: Review and Open Problems

AU - van Otterlo, M.

PY - 2002/7/9

Y1 - 2002/7/9

N2 - This paper is about representation in RL.We discuss some of the concepts in representation and generalization in reinforcement learning and argue for higher-order representations, instead of the commonly used propositional representations. The paper contains a small review of current reinforcement learning systems using higher-order representations, followed by a brief discussion. The paper ends with research directions and open problems.

AB - This paper is about representation in RL.We discuss some of the concepts in representation and generalization in reinforcement learning and argue for higher-order representations, instead of the commonly used propositional representations. The paper contains a small review of current reinforcement learning systems using higher-order representations, followed by a brief discussion. The paper ends with research directions and open problems.

KW - EWI-14945

KW - Review

KW - relational representations

KW - HMI-IA: Intelligent Agents

KW - IR-65334

KW - Generalization

KW - Reinforcement learning

M3 - Paper

T2 - ICML-2002 Workshop on Development of Representations

Y2 - 9 July 2002 through 9 July 2002

ER -

Relational Representations in Reinforcement Learning: Review and Open Problems

Abstract

Workshop

Keywords

Access to Document

Fingerprint

Cite this