基于免疫克隆选择的模糊聚类分析

Fuzzy cluster analysis based on immune clone selection

模糊c-均值聚类(FCM)应用广泛,但它容易陷入局部最优,且对初始值很敏感。提出了一种基于免疫克隆选择算法的模糊聚类方法,首先,用克隆选择算法对模糊聚类中心的个数和聚类中心的选取进行指导,然后,利用FCM进行聚类,是一种有监督学习和无监督学习结合的一种算法,实验结果表明:该方法在一定程度上避免FCM算法对初始值敏感和容易陷入局部最优解的缺陷,使聚类更有效,更合理。

The fuzzy c-means clustering（FCM） algorithm is a widely used clustering algorithm,but FCM has two shortcomings：the locality and the sensitivity. A new fuzzy clustering method based on the immune clone selection algorithm is proposed. First, the clone selection algorithm is used for guiding the fuzzy clustering center and the number of the fuzzy clustering center;then, FCM is used for clustering. That is an algorithm which is combined the supervised and the non-supervised. The experiment testifies that the method avoided the locality and the sensitivity of the FCM availably. This clustering method is more effective and reasonable.