Lenient Regret and Good-Action Identification in Gaussian Process Bandits

                   Xu Cai 1 Selwyn Gomes 1 Jonathan Scarlett 1 2

             Abstract                 gorithms can often be applied in a unified manner in these
                                  two settings.
   In this paper, we study the problem of Gaussian
   process (GP) bandits under relaxed optimization        Perhaps the most prominent class of existing results con-
   criteria stating that any function value abo ...