State Abstraction in MAXQ Hierarchical Reinforcement Learning

State Abstraction in MAXQ Hierarchical Reinforcement Learning (2000)

Cached

Download Links

[ftp.cs.orst.edu]
[web.engr.oregonstate.edu]

Other Repositories/Bibliography

DBLP

by Thomas Dietterich Department , Thomas G. Dietterich

Venue:

Advances in Neural Information Processing Systems 12

Versions

Version History

Metadata Version 3

User correction supplied by SystemCorrections

Datum	Value	Source
TITLE	State Abstraction in MAXQ Hierarchical Reinforcement Learning	user correction - Legacy Corrections
AUTHOR NAME	Thomas Dietterich Department	user correction - Legacy Corrections
AUTHOR NAME	Thomas G. Dietterich	user correction - Legacy Corrections
AUTHOR AFFIL	Department of Computer Science; Oregon State University; Corvallis, Oregon	user correction - Legacy Corrections
AUTHOR ADDR	97331-3202	user correction - Legacy Corrections
ABSTRACT	ion in MAXQ Hierarchical Reinforcement Learning Thomas G. Dietterich Department of Computer Science Oregon State University Corvallis, Oregon 97331-3202 tgd@cs.orst.edu Abstract Many researchers have explored methods for hierarchical reinforcement learning (RL) with temporal abstractions, in which abstract actions are defined that can perform many primitive actions before terminating. However, little is known about learning with state abstractions, in which aspects of the state space are ignored. In previous work, we developed the MAXQ method for hierarchical RL. In this paper, we define five conditions under which state abstraction can be combined with the MAXQ value function decomposition. We prove that the MAXQ-Q learning algorithm converges under these conditions and show experimentally that state abstraction is important for the successful application of MAXQ-Q learning. 1 Introduction Most work on hierarchical reinforcement learning has focused on temporal ab...	user correction - Legacy Corrections
YEAR	2000	INFERENCE
VENUE	Advances in Neural Information Processing Systems 12	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	994--1000	INFERENCE
VOLUME	12	INFERENCE
CITATIONS	12 found	ParsCit 1.0