Definition
Multi-agent learning (MAL) refers to settings in which multiple agents learn simultaneously. Usually defined in a game theoretic setting, specifically in repeated games or stochastic games, the key feature that distinguishes MAL from single-agent learning is that in the former the learning of one agent impacts the learning of others. As a result, neither the problem definition for multi-agent learning nor the algorithms offered follow in a straightforward way from the single-agent case. In this second of two entries on the subject, we focus on algorithms.
Some MAL Techniques
We will discuss three classes of techniques – one representative of work in game theory, one more typical of work in artificial intelligence (AI), and one that seems to have drawn equal attention from both communities.
Model-Based Approaches
The first approach to learning we discuss, which is common in the game theory literature, is the model-based one. It adopts the following general scheme:
-
...
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Requisite background in game theory can be obtained from the many introductory texts, and most compactly from Leyton-Brown and Shoham (2008). Game theoretic work on multiagent learning is covered in Fudenberg and Levine (1998) and Young (2004). An expanded discussion of the problems addressed under the header of MAL can be found in Shoham et al. (2007), and the responses to it in Vohra and Wellman (2007). Discussion of MAL algorithms, both traditional and more novel ones, can be found in the above references, as well as in Greenwald and Littman (2007).
Recommended Reading
Requisite background in game theory can be obtained from the many introductory texts, and most compactly from Leyton-Brown and Shoham (2008). Game theoretic work on multiagent learning is covered in Fudenberg and Levine (1998) and Young (2004). An expanded discussion of the problems addressed under the header of MAL can be found in Shoham et al. (2007), and the responses to it in Vohra and Wellman (2007). Discussion of MAL algorithms, both traditional and more novel ones, can be found in the above references, as well as in Greenwald and Littman (2007).
Fudenberg D, Levine D (1998) The theory of learning in games. MIT, Cambridge
Greenwald A, Littman ML (eds) (2007) Special issue on learning and computational game theory. Mach Learn 67(1–2):3–6
Leyton-Brown K, Shoham Y (2008) Essentials of game theory. Morgan and Claypool, San Rafael
Shoham Y, Powers WR, Grenager T (2007) If multiagent learning is the answer, what is the question? Artif Intell 171(1):365–377. Special issue on foundations of multiagent learning
Vohra R, Wellman MP (eds) (2007) Special issue on foundations of multiagent learning. Artif Intell 171(1)
Young HP (2004) Strategic learning and its limits. Oxford University Press, Oxford
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer Science+Business Media New York
About this entry
Cite this entry
Shoham, Y., Powers, R. (2017). Multi-agent Learning Algorithms. In: Sammut, C., Webb, G.I. (eds) Encyclopedia of Machine Learning and Data Mining. Springer, Boston, MA. https://doi.org/10.1007/978-1-4899-7687-1_569
Download citation
DOI: https://doi.org/10.1007/978-1-4899-7687-1_569
Published:
Publisher Name: Springer, Boston, MA
Print ISBN: 978-1-4899-7685-7
Online ISBN: 978-1-4899-7687-1
eBook Packages: Computer ScienceReference Module Computer Science and Engineering