基于粒子群和免疫克隆的模糊C-均值聚类

目的基于梯度下降的模糊聚类算法（FCM）选择最优解做改进,降低原FCM对初始值的敏感度,改进模糊C-均值算法,加快收敛速度,改善聚类的效果.方法该算法通过克隆选择改变粒子群优化算法（PSO）中群体的多样性,用PSO代替了FCM算法的基于梯度下降的迭代过程.结果算法具有很强的全局搜索能力,很大程度上避免了FCM算法易陷入局部极值的缺陷,同时也降低了FCM算法对初始值的敏感度.通过机器学习中的W ine和Iris数据对所提出的算法进行验证,取W ine样本数据为178个,条件属性为13,聚类类别数为3;Iris数据150个,条件属性个数为4,对这两类数据分别进行聚类分析,将试验结果与单纯的FCM和基于PSO的FCM比较,聚类的正确性有所提高.结论基于粒子群和免疫克隆的模糊C-均值聚类算法具有很强的全局搜索能力,提高了聚类的效果和效率.

Fuzzy C-means Clustering Based on the Particle Swam Optimization and Immune Clone

A new fuzzy cluster algorithm is proposed which takes the advantages of the particle swarm optimization with global searching and fast convergence and immune clonal selection algorithm with diversity antibodies. The new algorithm changes the diversity about particle swarm optimization through clone selection, and particle swarm optimization replaces the fuzzy C-mean iterative process based on the gradient descent method, which makes the new algorithm possess strong searching ability, avoids the fuzzy C-mean algorithm falling into local optimums and reduces the sensitivity to the initials of the fuzzy C-mean algorithm. A real application in classifying two data sets Wine and Iris in machine learning database is provided, Wine has 13 inputs,3 classes and 178 data vectors. Iris has 4 inputs,3 classes and 150 data vectors. The accuracy rate has been improved by the hybrid algorithm in comparison with FCM. Experiment results verify that global searching ability is strengthened by Fuzzy C-means clustering based on the particle swarm optimization and immune clone, efficiency and effectiveness of the presented algorithm is also improved.