摘要
提出一种基于粒子群算法的聚类算法,该算法利用粒子群算法随机搜索解空间的能力找到最优解.首先,将样本所属类号的组合作为粒子,构成种群,同时引入极小化误差平方和来指导种群进化的方向.其次,通过对全局极值的调整,搜索到全局最优值.最后,通过仿真实验的对比,验证了该算法在有效性和稳定性上要好于K-means算法.
A clustering method based on the particle swarm optimization is provided, using the ability of PSO algorithm which can search all of the solution space to find the optimum solution. Firstly, the combination of the cluster number of the samples was taken as particles to consist a swarm. Meanwhile, the evolution trend was used to modulate with the theory of the LMS error criterion. Secondly, according to the modulating for global best, the algorithm researched the global optimum. Finally, the simulation results show that the new algorithm of proposed algorithm is more efficient and stable than K-means algorithm.
出处
《延边大学学报（自然科学版）》
CAS
2009年第1期64-67,共4页
Journal of Yanbian University（Natural Science Edition）
关键词
粒子群
聚类
极小化误差平方和
particle swarm optimization
clustering
LMS error criterion
