Incentive Compatible Privacy-Preserving Distributed Classification
Source: University of Texas at Dallas
In this paper, the authors propose game-theoretic mechanisms to encourage truthful data sharing for distributed data mining. One proposed mechanism uses the classic Vickrey-Clarke-Groves (VCG) mechanism, and the other relies on the Shapley value. Neither relies on the ability to verify the data of the parties participating in the distributed data mining protocol. Instead, they incentivize truth telling based solely on the data mining result. This is especially useful for situations where privacy concerns prevent verification of the data. Under reasonable assumptions, they prove that these mechanisms are incentive compatible for distributed data mining. In addition, through extensive experimentation, they show that they are applicable in practice.