机构地区: 华南理工大学自动化科学与工程学院
出 处: 《哈尔滨工程大学学报》 2006年第B07期237-240,共4页
摘 要: 人工免疫系统是对基于生物免疫系统的模式识别机制而构造的信息处理方法的总称.对一种专门用于聚类分析的人工免疫系统模型——核聚类人工免疫网络进行研究.在核聚类人工免疫网络方法中,模式类的特征值的评价函数小仅是引导方法最终收敛的关键因素,同时,也决定了该方法的聚类分析效果.专门研究了核聚类人工免疫网络中特征值评价函数的具体形式及其对聚类分析效果的影响,从而为进一步完善和更好地应用该方法提供参考. Artificial immune system (AIS) is the collection of every kind of information processing methods constructed basing on the pattern identification mechanism of the natural immune system. In this paper, the research iis focused on the Kernel Clustering Artificial Immune Network (KCAIN), which is a kind of AIS specially designed for application in clustering analysis. The evaluation function of feature values of the classes in the KCAIN is not only a key factor for the convergence of this method, but also an important factor for the performance of the method. The effect of different forms of the evaluation function on results of clustering analysis is discussed in this paper, and the analysis of simulation results will provide references for application of the method.
领 域: [自动化与计算机技术] [自动化与计算机技术]