RKHS

Kernelized Bandits

Algorithm with uniformly improved regret bounds + a computationally efficient heuristic with better empirical performance