Impacts of team size on role learning in multiagent systems
Munetaka Saito · International conference on Artificial intelligence and applications · 2008
In multiagent systems, division of labor is essential for achieving tasks. To reduce the burden of the designer, it is preferable that agents assume their role by learning. Thus, it is important to clarify the appropriate conditions under which division of labor can easily emerge. In this paper, we focus on the impact of team size on role learning. We use a simple transportation task as the test problem and investigate the impact of the team size on the learnability of division of labor. The results show that a large team size is beneficial for learning of division of labor.