A policy is a decision rule for choosing an action from available information. In a multi-armed bandit, it selects which slot-machine arm to pull next while maximizing cumulative reward under limited trials and uncertain outcomes. section 5.1.1
Each choice balances exploitation—using a machine with a strong observed success rate—and exploration—testing machines whose reward probabilities remain uncertain. section 5.1.1
Choosing a slot machine
A player has a fixed number of pulls and several machines with unknown reward probabilities. The policy decides which arm to pull next, seeking the highest total number of coins. section 5.1.1
Figure 5.3 An example of an MAB dataset showing the exploration–exploitation dilemma. An MAB policy must choose between a machine with a consistently high success rate and one with a reward rate that is uncertain.
Compare a consistently successful machine with one whose reward rate is uncertain to see the exploration–exploitation dilemma. section 5.1.1
