一个基于博弈学习的多主体竞价模型  

A Dynamic Biding Price Model of Multi-Agents Based on Game Learning

作　　者：唐亮贵[1] 程代杰[1] 

机构地区：[1]重庆大学计算机学院

出　　处：《计算机工程与应用》2006年第17期76-78,104,共4页Computer Engineering and Applications

基　　金：重庆市自然科学基金资助项目(编号:CSTC;2004BB2167);重庆市教委科技项目

摘　　要：根据多主体撮合交易模型,把整个撮合交易看成各交易主体的动态交互过程,设计了基于Multi-Agent的电子商务交易市场中交易主体动态竞价策略,提出了博弈学习的概念,并建立了基于博弈学习的动态竞价模型,根据撮合密度的定义,分析了所建立模型的性能和效率,试验表明,基于博弈学习的多主体动态竞价模型使多主体撮合交易系统具有一定的自均衡和自学习能力和良好的交易性能。This paper regards the whole process of matchmaking tradeoff as each exchange agent dynamic alternation course based on multi-agent matchmaking tradeoff model,and designs the trade agent dynamic bid price strategy,puts forward the idea about the game learning,establishes the dynamic bid price model based on the game learning.By the definition of matchmaking density,this paper analyzes its performance and its efficiency.Experimentation indicates that, the biding price model of multi-agents by game learning can make multi-agent matchmaking tradeoff system self-equilibrium and self-learning,and have favorable dealing capability.

关 键 词：Muhi-Agent系统 撮合 动态竞标 博弈学习 

分 类 号：TP1[自动化与计算机技术—控制理论与控制工程]