eduzhai > Applied Sciences > Engineering >

Logical Team Q-learning An approach towards factored policies in cooperative MARL

  • Save

... pages left unread,continue reading

Document pages: 22 pages

Abstract: We address the challenge of learning factored policies in cooperative MARLscenarios. In particular, we consider the situation in which a team of agentscollaborates to optimize a common cost. The goal is to obtain factored policiesthat determine the individual behavior of each agent so that the resultingjoint policy is optimal. The main contribution of this work is the introductionof Logical Team Q-learning (LTQL). LTQL does not rely on assumptions about theenvironment and hence is generally applicable to any collaborative MARLscenario. We derive LTQL as a stochastic approximation to a dynamic programmingmethod we introduce in this work. We conclude the paper by providingexperiments (both in the tabular and deep settings) that illustrate the claims.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×