Deception in Game Theory: A Survey and Multiobjective Model
Deception in Game Theory: A Survey and Multiobjective Model
复制标题
博弈论中的欺骗：调查和多目标模型
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Austin Davis
中科院分区：
文献类型：
--
作者：
Austin Davis
博弈论是研究冲突的数学模型。它提供了用于分析多个代理之间的动态交互以及(在某些情况下)跨多个交互的工具。这篇论文由两篇学术文章组成，从博弈论(GT)的角度来研究欺骗。第一篇文章是对GT欺骗模型的综述。这项调查描述了研究人员如何使用博弈论来衡量欺骗的实用性，为实施欺骗的机制建模，分析欺骗的结果，以及应对或减轻欺骗的影响。这项调查突出了文献中的几个空白。一个重要的差距涉及在欺骗计划期间进行的利益-成本-风险权衡。为了弥补这一研究空白，第二篇文章介绍了一种对这些权衡进行建模的新方法。该方法使用GT欺骗模型来定义一个新的多目标优化问题，称为欺骗设计问题(DDP)。DDP的解决方案提供了欺骗行动的过程，这些过程在收益、成本和对欺诈者的风险方面都是有效的。一个基于空战模拟器输出的案例研究演示了7×7范式博弈中的DDP。在解决方案中观察到了两个突出的特征。首先，由此产生的许多解决方案都是零风险的。零风险解决方案包含以下两个属性之一：1。)如果欺骗被发现，被欺骗的玩家没有追索权，或者2。)除非被欺骗的玩家故意采用他认为不是最优的策略，否则欺骗是不可能被揭露的。其次，这些解决方案往往扭曲了游戏支出的相当大一部分：在98个支出中，至少有94个在七个有效的解决方案中每一个都被修改了。几个
Game theory is the study of mathematical models of conflict. It provides tools for analyzing dynamic interactions between multiple agents and (in some cases) across multiple interactions. This thesis consists of two scholarly articles that address deception from a game theoretic (GT) perspective. The first article is a survey of GT models of deception. The survey describes the ways researchers use game theory to measure the practicality of deception, model the mechanisms for performing deception, analyze the outcomes of deception, and respond to, or mitigate the effects of deception. The survey highlights several gaps in the literature. One important gap concerns the benefit-cost-risk trade-off made during deception planning. To address this research gap, the second article introduces a novel approach for modeling these trade-offs. The approach uses a GT model of deception to define a new multiobjective optimization problem called the deception design problem (DDP). Solutions to the DDP provide courses of deceptive action that are efficient in terms of their benefit, cost, and risk to the deceiver. A case study based on the output of an air-to-air combat simulator demonstrates the DDP in a 7× 7 normal form game. Two prominent features are observed in the solutions. First, many of the resulting solutions are zero-risk. A zero-risk solution implies one of two properties: either 1.) the deceived player has no recourse if the deception is discovered, or 2.) the deception cannot be revealed unless the deceived player intentionally adopts a strategy that he perceives to be suboptimal. Second, the solutions tended to distort a considerable portion of the game’s payouts: at least 94 of the 98 payouts were modified in each of the seven efficient solutions. Several