Multi-agent合作的委托-代理模型  

Principal-agent Model for Multi-agent Cooperation

作　　者：赵书良[1] 蒋国瑞[1] 黄梯云[1] 

机构地区：[1]北京工业大学经济与管理学院

出　　处：《计算机工程》2006年第22期46-48,106,共4页Computer Engineering

基　　金：北京市自然科学基金资助项目(9042001)

摘　　要：给出了Multi-agent合作的委托-代理模型,并利用该模型对常见的Multi-agent合作方式进行了分析,同时给出了模型的应用与讨论。模型通过委托人对代理人的最优激励合同设计,能在合作双方信息不对称情况下,促使代理人agent尽力工作,按委托人agent的意愿完成合作,从而能在一定程度上抑制合作中的欺骗,促进系统中agent间合作;同时,模型为合作伙伴达成合作付出代价的计算提供了理论依据。This paper gives a principal-agent model for multi-agent cooperation, and analyzes the usual agent cooperation methods with this model, finally gives examples and discusses of this model. Through the design of the incentive contract that the principal provides for the agent, this model can stimulate the agent to do its best for the principal in cooperation, and to accomplish the cooperation of the principal＇s will, even they are in the condition of asymmetric information about their cooperation. So this model can restrain the agent deception in cooperation, It also provides a method for calculating how much the cooperation initiator should pay for its cooperative partner for making the cooperation.

关 键 词：MULTI-AGENT 合作 委托-代理 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]