Reference
L. Buşoniu, D. Ernst, B. De Schutter, and R. Babuška,
"Continuous-state reinforcement learning with fuzzy approximation," in
Adaptive Agents and Multi-Agent Systems III. Adaptation and
Multi-Agent Learning (K. Tuyls, A. Nowé, Z. Guessoum, and D.
Kudenko, eds.), vol. 4865 of
Lecture Notes in Computer
Science, Berlin, Germany: Springer, ISBN 978-3-540-77947-6, pp. 27-43,
2008.
Abstract
Reinforcement Learning (RL) is a widely used learning paradigm for adaptive
agents. There exist several convergent and consistent RL algorithms which have
been intensively studied. In their original form, these algorithms require that
the environment states and agent actions take values in a relatively small
discrete set. Fuzzy representations for approximate, model-free RL have been
proposed in the literature for the more difficult case where the state-action
space is continuous. In this work, we propose a fuzzy approximation
architecture similar to those previously used for Q-learning, but we combine it
with the model-based Q-value iteration algorithm. We prove that the resulting
algorithm converges. We also give a modified, asynchronous variant of the
algorithm that converges at least as fast as the original version. An
illustrative simulation example is provided.
Publisher page
Downloads
BibTeX
@incollection{BusErn:07-030,
author = {Bu{\c{s}}oniu, Lucian and Ernst, Damien and De Schutter, Bart
and Babu{\v{s}}ka, Robert},
title = {Continuous-State Reinforcement Learning with Fuzzy
Approximation},
booktitle = {Adaptive Agents and Multi-Agent Systems III.\ Adaptation and
Multi-Agent Learning},
series = {Lecture Notes in Computer Science},
volume = {4865},
editor = {Tuyls, Karl and Now\'{e}, Ann and Guessoum, Zahia and Kudenko,
Daniel},
publisher = {Springer},
address = {Berlin, Germany},
pages = {27--43},
year = {2008}
}