Scaling Multi-Agent Reinforcement Learning with Selective Parameter Sharing

        Filippos Christianos 1 Georgios Papoudakis 1 Arrasy Rahman 1 Stefano V. Albrecht 1

              Abstract                (e.g. (Gupta et al., 2017)) whereby agents share some or all
                                  parameters in their policy networks. In the literature, pa-
   Sharing parameters in multi-agent deep reinforce-
                                  rameter sharing is typically applied indiscriminate ...