Bayesian Exploration: Incentivizing Exploration in Bayesian Games

In a wide range of recommendation systems, self-interested individuals (“agents”) make decisions over time, using information revealed by other agents in the past, and producing that may help future. Each agent would like to exploit best action given current but prefer previous explore various alternatives collect information. A social planner, means well-designed policy, can incentivize balance exploration exploitation order maximize welfare or some objective. The policy be modeled as multiarmed bandit algorithm under Bayesian incentivecompatibility (BIC) constraints. This line work has received considerable attention “economics computation” community. Although prior work, planner interacts with single at present paper allows affect one another directly shared environment. now face two sources uncertainty: what is environment, do? We focus on “explorable” actions: those recommended BIC policy. show how principal identify all such actions.