Abstract of FKI-166-92

Document-Name:  FKI-166-92 (not available via ftp)
Title:          Collective Learning and Action Coordination
Authors:	Gerhard Weiss
Revision-Date:  April 1992
Category:       Technical Report (Forschungsberichte Künstliche Intelligenz)
Abstract:       Learning in multi--agent systems establishes a young research 
                field in distributed artificial intelligence. This paper 
                investigates an action--oriented approach to delayed 
                reinforcement learning in reactive multi--agent systems, and 
                focusses on the question how the agents can learn to coordinate
                their actions. Two basic algorithms called the ACE algorithm 
                and the AGE algorithm (ACE and AGE stand for ``ACtion 
                Estimation'' and ``Action Group Estimation'', respectively) 
                for the collective learning of appropriate action sequences 
                are introduced. Both algorithms explicitly take into 
                consideration that the agents may know different aspects of 
                the environment and that actions may be incompatible.
                The experiments described in this paper illustrate these
                algorithms and their learning capacities.
Keywords:       Multi-agent systems, collective learning, action coordination
Size:           9 pages
Language:       English
ISSN:           0941-6358
Copyright:      The ``Forschungsberichte Künstliche Intelligenz''
                series includes primarily preliminary publications,
                specialized partial results, and supplementary
                material. In the interest of a subsequent final
                publication these reports should not be copied. All
                rights and the responsability for the contents of the
                report are with the authors, which would appreciate
                critical comments.



Gerhard Weiss