Feasible Joint Posterior Belief Distributions in Multi-Receiver Bayesian Persuasion
In multi-receiver Bayesian persuasion, while characterizing feasible joint belief distributions unconditionally is complex, conditioning on the state reveals a simple structure: feasibility constraints apply only to individual receivers' marginals across states, allowing arbitrary correlation within a state. This insight leads to a novel primal-dual approach for solving persuasion problems, leveraging optimal transport theory and offering tractable solutions for specific classes of problems.
커뮤니케이션에서 약속과 무작위성의 가치: 언제 약속이 발신자에게 유익한가?
유한한 행동과 상태를 가진 커뮤니케이션 게임에서 발신자에게 약속이 가치 있는 경우는 발신자가 무작위 전략을 선호하는 경우뿐이며, 이는 베이지안 설득 문제에서 부분적 실험이 아닌 무작위 실험이 최적일 때 발생한다.
Markov Persuasion Processes: Learning to Persuade from Scratch
Bayesian persuasion studies strategic information disclosure to influence behavior.
Markov Persuasion Processes: Learning to Persuade from Scratch
The authors introduce Markov persuasion processes (MPPs) to address sequential scenarios where a sender interacts with myopic receivers in an unknown environment. They propose an algorithm that optimizes regret and violation trade-offs in learning without knowledge of receiver rewards.
© 2024 by Linnk AI