Offline Reinforcement Learning with Pseudometric Learning

Robert Dadashi 1 Shideh Rezaeifar 2 Nino Vieillard 1 3 Leonard Hussenot 1 4 Olivier Pietquin 1 Matthieu Geist 1

              Abstract               that generated these experiences (Pomerleau, 1991). How-
                                ever, if these experiences come from different sources, with
    Offline Reinforcement Learning methods seek
                                different degrees of desirability, naive imitation might lea ...