AN EMPIRICAL STUDY OF POTENTIAL-BASED REWARD SHAPING AND ADVICE IN COMPLEX, MULTI-AGENT SYSTEMS
Advances in Complex Systems · 2011-04 · 14 (02) · 251-278
DOI: 10.1142/s0219525911002998 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advances in Complex Systems, online erschienen am 20.11.2011. Verfasst von SAM DEVLIN und 2 weiteren Autor:innen. Die Arbeit wurde bisher 39-mal zitiert (Crossref) und nennt 19 Literaturangaben.
Abstract (Originalsprache)
This paper investigates the impact of reward shaping in multi-agent reinforcement learning as a way to incorporate domain knowledge about good strategies. In theory, potential-based reward shaping does not alter the Nash Equilibria of a stochastic game, only the exploration of the shaped agent. We demonstrate empirically the performance of reward shaping in two problem domains within the context of RoboCup KeepAway by designing three reward shaping schemes, encouraging specific behaviour such as keeping a minimum distance from other players on the same team and taking on specific roles. The results illustrate that reward shaping with multiple, simultaneous learning agents can reduce the time needed to learn a suitable policy and can alter the final group performance.
Zitieren
APA
DEVLIN, S., KUDENKO, D., & GRZEŚ, M. (2011). AN EMPIRICAL STUDY OF POTENTIAL-BASED REWARD SHAPING AND ADVICE IN COMPLEX, MULTI-AGENT SYSTEMS. Advances in Complex Systems, 14(02), 251-278. https://doi.org/10.1142/s0219525911002998
Vancouver
DEVLIN S, KUDENKO D, GRZEŚ M. AN EMPIRICAL STUDY OF POTENTIAL-BASED REWARD SHAPING AND ADVICE IN COMPLEX, MULTI-AGENT SYSTEMS. Advances in Complex Systems. 2011;14(02):251-278. doi:10.1142/s0219525911002998
BibTeX
@article{DEVLIN2011,
author = {DEVLIN, SAM and KUDENKO, DANIEL and GRZEŚ, MAREK},
title = {AN EMPIRICAL STUDY OF POTENTIAL-BASED REWARD SHAPING AND ADVICE IN COMPLEX, MULTI-AGENT SYSTEMS},
journal = {Advances in Complex Systems},
year = {2011},
volume = {14},
number = {02},
pages = {251--278},
doi = {10.1142/s0219525911002998}
}
EUVIMED