【作者】 刘文俊；

【导师】 王天江；

【作者基本信息】 华中科技大学 ， 计算机应用技术， 2013， 博士

【摘要】 协商是多方合作中分布式冲突消解的重要机制。随着分布式应用的发展,环境的不确定性和动态性导致应用越来越复杂,从而需要设计具有针对性的协商处理机制。在此基础上,分布式问题求解越来越关注于个体问题求解规划的自治性、个体利益和共同利益达成间的权衡,并在保留必要交互的同时确保个体的隐私。这需要在基于私有和不确定信息的分布式环境中存在一种有效的群体规划机制来约束和管理智能体的问题求解和协商能力。另外,多媒体数据库中的相似性搜索的精确度和效率的改进近来往往求助于多个度量的联合使用。由于(特别是非线性)多度量动态搜索的复杂性,至今还未有较好的通用机制来对其进行处理。出于通用性考虑,将多度量动态搜索视为一种分布式规划求解过程；则群体规划在此应用上的潜在价值很值得我们进行关注。本文的主要工作及创新点如下：(1)在研究不确定动态条件下双边协商的基础上,本文提出了一个基于外部竞争和选择考虑的动态双边协商决策模型——COBN模型。该模型在时间约束下适用于信息不完全的场景,它描述了多对多关系下的价格协商。在分析来自外部选择的最大期望收益的过程中,考虑外部竞争对协商对手保留收益的影响,权衡外部选择对应协商达成的可能性,调整本方的保留收益和协商报价,在竞争与选择并存的不确定动态环境下谋取最佳收益。通过与以往双边协商决策机制的对比实验分析和解释,验证了本算法能够在不确定动态环境下获得更好的协商效果,并在模型参数与中间算法的选择上取得了初步的经验,为进一步的研究工作奠定了基础。(2)作为多智能体问题求解的有效手段,必需的自治个体规划和交互由于适应其他个体利益的能力有限,可能对智能体解决问题和协商的能力产生不必要制约,特别是在基于私有信息和不确定信息的分布式协作环境中。本文以协同进化论为理论依据,首次提出了多智能体群体规划模型。其中,每个智能体通过进化搜索执行局部规划,并依据协同进化机制消解冲突,通过聚类和群体选择的方式在交互过程中修正各自的搜索方向,从而达成局部和全局目标的优化。该模型基于协同进化的社会流程将在线聚类分析和群体选择集成进分布式规划中；并设计了有针对性的在线聚类和交互迁移机制,以改善交互的效率、避免通信拥塞,确保群体规划的关键——分布式动态环境下的异步、自适应、协作和自主性。(3)面向一般(尤其是非线性)多度量动态搜索提出了一个基于群体规划的通用模型。该模型以多度量小世界网为底层支撑,在局部搜索中引入梯度学习和协同学习策略改进规划效率,根据分布式搜索的特点设计了具有针对性的聚类分析和交互迁移机制,并因地制宜地简化了群体协调机制,为多度量动态搜索量身定做了一套全新的群体规划流程。该流程满足了多度量动态搜索的鲁棒性、不确定动态性及通用性要求。更多还原

【Abstract】 As an effective means of multi-agent problem solving, autonomous individual programming and interaction that are essential need, are limited in their ability to accommodate the interests of others, and therefore, may unnecessarily constrain the solving ability and negotiability of an agent, particularly in a distributed cooperative environments founded on private and uncertain information. In this paper, a multi-agent group programming model is presented, where each agent executes local programming by evolutionary search. Based on co-evolution idea, agents resolve conflicts and revise their own search direction to optimize local and social objectives in an interactive process by means of clustering and group choice.Moreover, a recent trend to improve the e□ectiveness of the similarity search in multi-media database resorts to use several metric functions. For the sake of complexity of (especially nonlinear) dynamic similarity search in multi-metric spaces, so far there are no general mechanisms in this domain. For general considerations, dynamic similarity search in multi-metric spaces can be viewed as a kind of distributed programming for problem-solving. Applications of group programming in the domain are worthy of our attentions.The main work and innovations are as follows:(1) Based on the research of bilateral negotiations with uncertain dynamics, a new model, COBN (competitions and options based negotiation), is presented. With limitation of time, this model is suitable for environment with incomplete information, and describesprice negotiations in many-to-many relationshuip.(2) A multi-agent group programming model is presented, where each agent executes local programming by evolutionary search. Based on co-evolution idea, agents resolve conflicts and revise their own search direction to optimize local and social objectives in an interactive process by means of clustering and group choice.(3) For the need of (especially nonlinear) dynamic similarity search in multi-metric spaces, a general model based on group programming is presented, where the base index infrastructure is a multi-metric small world network. Gradient learning and cooperative learning strategy are introduced to group program on the query object, in order to meet the need of robustness, uncertainty, dynamic and generality in dynamic similarity search in multi-metric spaces.更多还原