摘要： 介绍了基于信息熵的免疫遗传算法的聚类分析方法。将免疫算法引入到遗传算法中，利用免疫算法的免疫记忆、自我调节和多样性保持功能弥补了标准遗传算法的局部搜索能力差、计算量大和早熟收敛等问题。采用DNA进行抗体编码，利用信息熵来表示抗体间亲和度及浓度，并采用聚合亲和度，实现了抗体群的自我调节和多样性保持策略。实验表明，该算法优于标准遗传算法。

关键词: 聚类, 免疫遗传算法, 信息熵, 亲和度

Abstract: This paper presents a clustering analysis method of immune genetic algorithm based on entropy. Immune algorithm is introduced into genetic algorithm toincorporate immune algorithm’s functions such as immune memory, self-adjustability and the keeping of the diversity. The immune-genetic algorithm based on information entropy can overcome the shortcomings of standard genetic algorithm, e. g., poor local search capability, excessive computational cost and premature convergence, etc. This algorithm adopts DNA to code the antibody, and uses information entropy to denote the affinity and the consistence of the antibodies. The converged affinity is presented to achieve self-adjustability of colony, which keeps the diversity of colony. The experiment demonstrates that the algorithm is better than standard genetic algorithm in clustering analysis.

Key words: clustering, immune-genetic algorithm, information entropy, affinity

中图分类号: 

• TP311