摘要： 利用增强学习的Q-learning理论，构造了一个基于时间信念、价格信念和状态期望Q值的代理谈判模型。将代理交互报价的过程看成代理选择报价行动，实现状态的迁移，可以计算出代理在不同状态采取行动的Q值。代理可以通过修改信念函数及时间贴现率来调整报价。该文实现了谈判模型的报价算法，并从理论和实验数据两方面进行了分析比较。

关键词: 代理, 增强学习, 报价, 时间信念, 价格信念

Abstract: An agent negotiation model based on enforcement learning, time belief, price belief and state expected Q value is developed. The alternative pricing process of agents is considered as state transition, and the Q value of the state-action pair can be computed. By modifying belief functions and time discount rate, agent can adjust its pricing. The algorithm of negotiation pricing is implemented.

中图分类号: 

• TP181