Kernelized Bandits

We study the kernelized bandit problem, that involves designing an adpative strategy for querying a noisy zeroth-order-oracle to efficiently learn about the optimizer of an unknow function $f$ with a norm bounded by $M$ in a Reproducing kernel …