A Study of Reducing Reveled Information in Decentralized Asymmetric Constraint Optimization

Abstract

In cooperative problem solving and negotiation of multiple agents, reduction of revealed information from agents is an important problem. In this study, we focus on decentralized asymmetric constraint optimization methods as a fundamental negotiation framework reducing revealed information among agents who have individual private objective function values for the decision of their related agents. 1) A goal of the study is to reduce the information of objective values that are published and revealed from agents in the solution process for optimizing the objectives of individual agents. To this end, we present a heuristic solution framework and related criteria for an agreement of agents on a solution that is a compromise between revealed information and solution quality. 2) As a result of the study, we experimentally show effect and influence of several heuristic methods that consider different criteria to select published information of agents. With the result, we discuss future directions of the study.