A Study of Cooperation with Privacy Loss Based on Asymmetric Constraint Optimization Problem Among Agents
Toshihiro Matsui · 2019
Privacy preservation is one motivation underlying cooperative optimization problems among agents. Although several recent solution methods for constraint optimization problems for agents have employed secure computation to completely hide the privacy information of agents, some information might need to be published to manage the solution by a mediator or to review the results with aggregated information. We focus on asymmetric constraint optimization problems among agents whose utilities are differently defined as asymmetric constraints. To obtain solutions, agents should reveal some part of their constraints, although the amount of the revealed evaluation values is considered privacy loss. We define a framework of such solution processes and investigate several fundamental heuristic strategies of agents.