|
|
Learning to take turns Host Publication: Finds and Results from the Swedish Cyprus Expedition: A Gender Perspective at the Medelhavsmuseet Authors: P. Vrancx, K. Verbeeck and A. Nowé Publication Year: 2010 Number of Pages: 7
Abstract: This paper provides a novel approach to multi-agent coordination in general sum Markov games. Contrary to what is common in multi-agent learning, our approach does not focus on reaching a particular equilibrium between agent policies. Instead, it learns a basis set of special joint agent policies, over which it can randomize to build different solutions.
The main idea is to tackle a Markov game by decomposing it into a set of multi-agent common interest problems each reflecting one agent's preferences in the system. With only a minimum of coordination, simple reinforcement learning agents using Parameterised Learning Automata are able to solve this set of MMDPs in parallel.
A third party is used to select the MMDP to be played, without a need for the agents to know which game or re- ward function they are confronted with. As a result, a team of simple learning agents becomes able to switch play be- tween desired joint policies rather than mixing individual policies. One application of this principle, which we consider in this paper, is to let simple adaptive agents learn to take turns in general sum Markov Games in order to satisfy their individual objectives. We experimentally demonstrate this principle in a grid-world setting
|
|
