GP-UCB

Regret Optimality of GP-UCB