arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1709.01432cs.GT
收藏

有意见交换的联盟博弈

Coalitional game with opinion exchange

Bomin Jiang, Mardavij Roozbehani, Munther A. Dahleh

摘要:

在联盟博弈中,如果不同的参与者对联盟中每个子集的支付函数持有不同的观点,传统的联盟博弈理论就不适用了。在本文中,我们提出了一个框架,在这个框架中,参与者可以交换他们对支付函数的看法,然后决定大联盟的价值分配。当所有参与者都说真话时,意见一致的问题与联盟博弈无关,但当参与者在共识阶段具有战略性时,就会出现有趣的动态。假设所有参与者都是理性的,该模型意味着,如果有影响力的参与者是风险厌恶的,则在纯策略纳什均衡实现分布式数据的有效融合,这意味着平均意见不会漂移。此外,在不假设所有参与者都是理性的情况下,每个参与者都可以使用算法R学习过程,这与理性参与者的纯策略纳什均衡的结果相同。

英文摘要:

In coalitional games, traditional coalitional game theory does not apply if different participants hold different opinions about the payoff function that corresponds to each subset of the coalition. In this paper, we propose a framework in which players can exchange opinions about their views of payoff functions and then decide the distribution of the value of the grand coalition. When all players are truth-telling, the problem of opinion consensus is decoupled from the coalitional game, but interesting dynamics will arise when players are strategic in the consensus phase. Assuming that all players are rational, the model implies that, if influential players are risk-averse, an efficient fusion of the distributed data is achieved at pure strategy Nash equilibrium, meaning that the average opinion will not drift. Also, without the assumption that all players are rational, each player can use an algorithmic R-learning process, which gives the same result as the pure strategy Nash equilibrium with rational players.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑