Neural fitted Q iteration - First experiences with a data efficient neural reinforcement learning method

Autor(en): Riedmiller, M
Herausgeber: Gama, J
Camacho, R
Brazdil, P
Jorge, A
Torgo, L
Stichwörter: Computer Science; Computer Science, Artificial Intelligence
Erscheinungsdatum: 2005
Herausgeber: SPRINGER-VERLAG BERLIN
Journal: MACHINE LEARNING: ECML 2005, PROCEEDINGS
LECTURE NOTES IN ARTIFICIAL INTELLIGENCE
Volumen: 3720
Startseite: 317
Seitenende: 328
Zusammenfassung: 
This paper introduces NFQ, an algorithm for efficient and effective training of a Q-value function represented by a multi-layer perceptron. Based on the principle of storing and reusing transition experiences, a model-free, neural network based Reinforcement Learning algorithm is proposed. The method is evaluated on three benchmark problems. It is shown empirically, that reasonably few interactions with the plant are needed to generate control policies of high quality.
Beschreibung: 
16th European Conference on Machine Learning (ECML)/9th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD), Oporto, PORTUGAL, OCT 03-07, 2005
ISBN: 9783540292432
ISSN: 03029743

Show full item record

Page view(s)

1
Last Week
0
Last month
1
checked on Mar 3, 2024

Google ScholarTM

Check

Altmetric