一种基于多智能体策略的人工免疫网络数据分类方法

发布时间：2018-04-08 12:30

本文选题：人工免疫网络　切入点：多智能体技术　出处：《计算机应用研究》2017年01期

【摘要】：在传统的人工免疫网络基础上,将多智能体技术的典型策略融入到免疫网络的进化过程中。算法引入了邻域克隆选择,操作过程从局部到整体,能够更加全面地模拟免疫网络的自然进化模型;同时在免疫网络进化过程中增加了抗体间的竞争和协作操作,提高了网络的动态分析能力。后续实验中,分别采用常用的三组UCI数据和一幅红树林多光谱TM遥感图像对算法加以验证,实验结果表明算法对遥感图像有较高的分类效率,对UCI数据也有较好的分类效果,表明该算法是一种有效的数据分类方法。
[Abstract]:Based on the traditional artificial immune network, the typical strategy of multi-agent technology is integrated into the evolution of immune network.The algorithm introduces neighborhood clone selection, the operation process is from local to whole, it can simulate the natural evolution model of immune network more comprehensively, at the same time, the competition and cooperative operation among antibodies are increased in the evolution process of immune network.The dynamic analysis ability of the network is improved.In subsequent experiments, three groups of UCI data and a mangrove multispectral TM remote sensing image are used to verify the algorithm. The experimental results show that the algorithm has higher classification efficiency for remote sensing images and better classification effect for UCI data.It is shown that this algorithm is an effective method for data classification.
【作者单位】：华侨大学工学院;华侨大学计算机学院;
【基金】：华侨大学研究生科研创新能力培育计划资助项目
【分类号】：TP18;TP751

【相似文献】