期刊文献+

Agent的行为自信度及其修改规则 被引量:6

SELF-CONFIDENCE DEGREE OF AGENT'S BEHAVIORS AND ITS REVISION RULES
在线阅读 下载PDF
导出
摘要 在多Agent系统中,个体Agent除了具有自治性、应激性、主动性外,还应当具有社会性以完成只有协作才能解决的问题.这种社会性要求Agent能及时响应其它Agent的服务请求,对自己的行为作出承诺.但是,有能力和意图并不一定有信心.本文从Agent的社会行为出发,为A-gent引入行为自信度的概念,并提出了理性Agent应当遵守的行为自信度修改规则.最后给出了一个动作委托算法,为设计具体的多Agent系统提供了一个可行的解决方案. In multi-agent systems, individual agents should also have social abili-ties in order to complete their own problem solving and to help others with their ac-tivities, as well as autonomy, responsiveness and proactiveness. Social ability re-quires that agents should respond timely to requests from other agents and committo their own actions. If an agent advertises a willingness to perform a service, thenit is obliged to perform this service when asked to do so. However, ability and in-tention don't mean confidence. An agent cannot be sure whether the services per-formed satisfy its clients or not. Especially when more than one agent has the samecapabilities to provide the services, all of them contest for the opportunity, whichowned by one, all the others lost. Knowing that when it fails, the client will neverget satisfying services which it should have, no matter whether this agent has e-nough confidence to per form successfully. So, action entrusting should be done ac-cording to certain rules or strategies. From social behaviors of agents, the conceptof self-confidence degree of agent's behaviors is introduced in this paper. Some mo-dification rules for self-confidence degree of behaviors which agents should meetin multi-agent systems are also characterized. Finally an action entrusting algo-rithm is designed, which provides a feasible and reasonable solution for the designof actual multi-agent systems.
出处 《计算机学报》 EI CSCD 北大核心 1998年第1期40-46,共7页 Chinese Journal of Computers
基金 国家863高科技基金 国家自然科学基金
关键词 多AGENT系统 行为自信度 动作集 人工智能 Multi-agent system, self-confidence degree of agent's behaviors, action set
  • 相关文献

同被引文献48

引证文献6

二级引证文献23

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部