Function Contrastive Learning of Transferable Meta-Representations

    Muhammad Waleed Gondal 1 Shruti Joshi 1 Nasim Rahaman 1 2 Stefan Bauer 1 3 Manuel Wuthrich 1
                     Bernhard Scholkopf 1

              Abstract                 rely on the number of samples N being large. This poses
    Meta-learning algorithms adapt quickly to new        further challenges: in many domains of interest, sourcing
    tasks that are drawn from the same task distribu-      enough data is a chal ...