论文部分内容阅读
AODE是我们研制的一个面向agent的智能系统开发环境,本文以AOD为平台研究了多agent环境下的协商与学习,本文利用协商-协商过程-协商线程的概念建立了多边-多问题协商模型MMN,该协商模型支持多agent环境中的多种协商形式及agent在协商过程中的学习,系统中的学习agent采用状态概率聚类空间上的多agent强化学习算法,该算法通过使用状态聚类方法减少Q值表存储所需空间,降低了经典Q-学习算法由于使用Q值表导致的对系统计算资源的要求,且该算法仍然可以保证收敛到最优解。
AODE is an agent-oriented intelligent system development environment that we developed. This paper studies the negotiation and learning under multi-agent environment based on AOD. This paper uses the concept of negotiation-negotiation process-negotiation thread to establish the MMN , The negotiation model supports multiple negotiation forms in multi-agent environment and agent learning in the negotiation process. The learning agent in the system adopts the state probability clustering multi-agent reinforcement learning algorithm in space, which uses state clustering method Reducing the space required for Q-value table storage reduces the computational resource requirements of classical Q-learning algorithm due to the use of Q-value tables, and the algorithm can still guarantee convergence to the optimal solution.