A Truthful Budget Feasible Multi-Armed Bandit Mechanism for Crowdsourcing Time Critical Tasks

A Truthful Budget Feasible Multi-Armed Bandit Mechanism for Crowdsourcing Time Critical Tasks
复制标题

DOI:
--
复制
发表时间:
2015-05
期刊:
--
影响因子:
--
通讯作者:
Arpita Biswas;Shweta Jain;Debmalya Mandal;Y. Narahari
Arpita Biswas;Shweta Jain;Debmalya Mandal;Y. Narahari
中科院分区:
其他
文献类型:
--
作者:
Arpita Biswas;Shweta Jain;Debmalya Mandal;Y. Narahari

文献摘要

被引文献

相似文献

基于现代众包平台上服务请求者面临的分配和定价问题,研究了一个具有以下特征的多臂强盗(MAB)问题:(a)请求者希望众包多个任务,但有固定的预算,这导致在分配任务时需要权衡成本和质量;(B)每个任务都有一个固定的截止日期,并且分配给任务的工作人员在该截止日期之前是不可用的;(c)拥挤工作人员的质量(在截止日期内成功完成任务的概率)是未知的;以及(d)拥挤工作人员对他们的成本是有策略的。我们提出了一种机制,最大限度地提高预期数量的成功完成的任务,确保预算的可行性,激励相容性,和个人的理性。我们建立了一个上限为O(B2/3(Kln(KB))1/3)的预期遗憾的建议机制相对于一个适当的基准算法,其中B是总预算和K是工人的数量。接下来,我们提供了任何确定性的真实机制,解决上述问题的一个特征,并使用此特征建立一个Ω的下限(B2= 3 K1 =3)的预期遗憾的任何预算的MAB机制满足上述性质。
Motivated by allocation and pricing problems faced by service requesters on modern crowdsourcing platforms, we study a multi-armed bandit (MAB) problem with several real-world features: (a) the requester wishes to crowdsource a number of tasks but has a fixed budget which leads to a trade-off between cost and quality while allocating tasks to workers; (b) each task has a fixed deadline and a worker who is allocated a task is not available until this deadline; (c) the qualities (probability of completing a task successfully within deadline) of crowd workers are not known; and (d) the crowd workers are strategic about their costs. We propose a mechanism that maximizes the expected number of successfully completed tasks, assuring budget feasibility, incentive compatibility, and individual rationality. We establish an upper bound of O(B2/3(K ln(KB))1/3) on the expected regret of the proposed mechanism with respect to an appropriate benchmark algorithm, where B is the total budget and K is the number of workers. Next, we provide a characterization of any deterministic truthful mechanism that solves the above class of problems and use this characterization to establish a lower bound of Ω(B2=3K1=3) on the expected regret for any budgeted MAB mechanism satisfying the above properties.