How does a gambler maximize winnings from a row of slot machines? This is the inspiration for the "multi-armed bandit problem," a common task in reinforcement learning in which "agents" make choices ...