A Natural Solution Explore-then-Exploit(ETE): (1)Do explore for the first To round by pulling each arm for To/K times; (2)Do exploit for the rest T-To round by always pulling a=arg maxael]r(a). Theorem1.Suppose that∀t∈[T]anda∈[K],0≤rt(a)≤l,then ETE with exploration period To guarantees eets(段+Tep()a a∈[K1 Advanced Optimization(Fall 2023) Lecture 12.Stochastic Bandits 7Advanced Optimization (Fall 2023) Lecture 12. Stochastic Bandits 7 A Natural Solution • Explore-then-Exploit (ETE):