基于PSO的可能性C均值聚类算法的研究

Possibilistic C-Means Clustering Algorithm Based on Particle Swarm Optimization

可能性C均值算法(PCM)是为了克服模糊C均值算法对噪声的敏感性而提出来的,但是它也存在一些缺陷,如易陷入局部最优,对初始条件敏感,导致聚类结果一致性等问题。针对以上问题,通过引进粒子群算法对其进行改进可以有效地避免这些问题,即提出了基于粒子群优化的可能性C均值聚类算法(PSO-PCM)。基于粒子群优化的可能性C均值聚类方法首先对编码过的数据点进行优化,然后对该方法产生的中心点进行聚类,在聚类的过程中根据适应度函数再进行调节。通过对给定数据集的聚类测试,结果表明,基于粒子群优化的可能性C均值聚类方法在收敛速度和全局寻优能力等方面有较大的改进。

The Possibility of C-means algorithm（PCM） is proposed to overcome the sensitivity of fuzzy C-means algorithm to noises. However,it also has some defects,such as easily to fall into local optimum,sensitivity to initial conditions and leading to consistency of clustering results. For the above problems,the particle swarm optimization algorithm can be improved to avoid them for possibilistic C-means algorithm,which is named PSOPCM. The new algorithm optimizes the encoded data points,and clusters the center points. And then,according to the fitness function,the center points are adjusted in the process of clustering. Through testing the given data set,the results show that PSOPCM algorithm has a greater improvement in convergence speed and global optimization ability.