中文摘要：

以人工免疫网络理论结合k-平均算法,尝试了一种聚类分析的新的解决方案.对k-平均算法中每一次迭代求平均值来确定聚类中心的方式进行改进,采用人工免疫网络中克隆选择和变异机制对聚类中心进行操作,选取最优抗体作为下一次迭代的聚类中心,克服了k-平均算法中对孤立点敏感的缺点,从而大大减少了迭代次数.通过对4组标准数据的实验,结果表明,该算法具有很好的自适应性,收敛速度快,提高了聚类性能.

英文摘要：

A novel solution for k-means cluster based on artificial immune net is presented. To achieve cluster centers in each iteration of k-means, we used clone selection and variation in artificial immune net, instead of taking mean value, as cluster center. In this new approach, best antibodies were selected as cluster centers of next iteration, so that the sensitivity for isolated point consisting of k-means algorithm was overcame, therefore the iteration process is reduced. This algorithm was tested on four groups of standard data sets, and was found to have good self-adaption and performance.