当前位置: X-MOL 学术Auton. Agent. Multi-Agent Syst. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A contract-based incentive mechanism for distributed meeting scheduling: Can agents who value privacy tell the truth?
Autonomous Agents and Multi-Agent Systems ( IF 2.0 ) Pub Date : 2021-07-02 , DOI: 10.1007/s10458-021-09516-8
Boya Di 1 , Nicholas R. Jennings 2
Affiliation  

We consider a distributed meeting scheduling problem where agents negotiate with each other to reach a consensus over the starting time of the meeting. Each agent has a private preference over a set of time slots, and aims to select its own preferred slot while revealing as little information about its preference as possible. A key challenge in this canonical setting is whether it is possible to design a distributed mechanism where agents that value their privacy are motivated to tell the truth about their preferences. In this paper, we give a positive answer by proposing a novel incentive mechanism based on economic contract theory. A set of contracts are carefully designed for agents of different types, consisting of the required actions, corresponding rewards and the privacy leakage level. By selecting the contract that maximises its own utility, each agent will not deviate from the required actions and can avoid unnecessary privacy leakage. Other properties of the mechanism such as budget balance, no need for a central authority, and near-optimal social welfare are also theoretically proved. Our empirical evaluations show that our proposed mechanism reduces privacy leakage by 58% compared to a standard calendar-sharing scheme. The social welfare of the proposed mechanism reaches over 88% of the optimal centralized method, and is higher than the social welfare of the state-of-the-art schemes by between 16 and 82%. A better trade-off between the privacy leakage and the number of rounds for convergence is also achieved compared to a typical negotiation mechanism.



中文翻译:

一种基于契约的分布式会议调度激励机制:重视隐私的座席能不能说真话?

我们考虑分布式会议安排问题,其中代理相互协商以在会议开始时间上达成共识。每个代理对一组时隙都有一个私人偏好,旨在选择自己的首选时隙,同时尽可能少地透露有关其偏好的信息。这种规范设置中的一个关键挑战是,是否有可能设计一种分布式机制,让重视隐私的代理有动力说出他们的偏好真相。在本文中,我们通过提出一种基于经济契约理论的新型激励机制给出了肯定的答案。为不同类型的代理精心设计了一套合约,包括所需的动作、相应的奖励和隐私泄漏级别。通过选择最大化自身效用的合约,每个代理都不会偏离所需的动作,可以避免不必要的隐私泄露。该机制的其他特性,如预算平衡、不需要中央权威和接近最优的社会福利,也在理论上得到了证明。我们的实证评估表明,与标准日历共享方案相比,我们提出的机制将隐私泄露减少了 58%。所提出机制的社会福利达到最优集中方法的 88% 以上,比最先进方案的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。每个代理都不会偏离所需的动作,可以避免不必要的隐私泄露。该机制的其他特性,如预算平衡、不需要中央权威和接近最优的社会福利,也在理论上得到了证明。我们的实证评估表明,与标准日历共享方案相比,我们提出的机制将隐私泄露减少了 58%。所提出机制的社会福利达到最优集中方法的 88% 以上,比最先进方案的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。每个代理都不会偏离所需的动作,可以避免不必要的隐私泄露。该机制的其他特性,如预算平衡、不需要中央权威和接近最优的社会福利,也在理论上得到了证明。我们的实证评估表明,与标准日历共享方案相比,我们提出的机制将隐私泄露减少了 58%。所提出机制的社会福利达到最优集中方法的 88% 以上,比最先进方案的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。理论上也证明了接近最优的社会福利。我们的实证评估表明,与标准日历共享方案相比,我们提出的机制将隐私泄露减少了 58%。所提出机制的社会福利达到最优集中方法的 88% 以上,比最先进方案的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。理论上也证明了接近最优的社会福利。我们的实证评估表明,与标准日历共享方案相比,我们提出的机制将隐私泄漏减少了 58%。所提出机制的社会福利达到最优集中方法的 88% 以上,比最先进方案的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。并且比最先进计划的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。并且比最先进计划的社会福利高 16% 至 82%。与典型的协商机制相比,还实现了隐私泄漏和收敛轮数之间的更好权衡。

更新日期:2021-07-02
down
wechat
bug