Approximation Theory Based Methods for RKHS Bandits
Sho Takemori 1 Masahiro Sato 1
Abstract the adversarial RKHS bandit problem, where a learner inter-
The RKHS bandit problem (also called kernel- acts with a sequence of any functions from the RKHS with
ized multi-armed bandit problem) is an online op- bounded norms. However, they only consider the kernel
timization problem of non-linear functions with loss, i.e., a ...


雷达卡




京公网安备 11010802022788号







