Explanation:

Stochastic multi-armed bandit (MAB) has been extensively studied in machine learning and sequential decision making. The most simple version of this problem consists of K arms, where each arm has an unknown distribution of the reward.