基于人工蜂群与K-Means的改进混合聚类算法

Improved Hybrid Clustering Algorithm Based on Artificial Bee Colony Algorithm and K-Means Algorithm

摘要: 为了克服K-Means聚类算法过度依赖初始聚类中心、容易陷入局部最优的缺点以及人工蜂群算法因为搜索策略的局限而导致的易早熟，收敛速度慢的问题，提出了改进的全局人工蜂群算法与K-Means++算法相结合的混合聚类方法，充分利用改进的全局人工蜂群算法可以全局寻优的特点与K-Means++算法能够优化初始聚类中心位置并且收敛速度快的特点，将二者融合，使得K-Means可以进行全局搜索，跳出局部最优解，并用UCI数据库中的Wine数据集和Balance-Scale数据集进行实验。结果表明，改进的全局人工蜂群算法较标准人工蜂群算法收敛速度更快，寻优效果更好；本文提出的混合聚类算法与原始K-Means算法相比，稳定性更好，迭代次数减少，收敛速度更快，而且聚类效果也有了明显改善。

Abstract: In order to overcome the disadvantages of K-Means clustering algorithm, such as over dependence on the initial clustering center, easily falling into local optimum, and the premature and slow con-vergence of the artificial bee colony algorithm due to the limitations of search strategies, a hybrid clustering method combining the improved global artificial bee colony algorithm and K-Means++ algorithm is proposed, which makes full use of the characteristics of the improved global artificial bee colony algorithm and K-Means++ algorithm. It can optimize the location of the initial clustering center and the convergence speed is fast. By combining the two, K-Means can search globally and jump out of the local optimal solution. The experiments are carried out with the Wine data set and balance-Scale data set in the UCI database. The results show that the improved global artificial bee colony algorithm has faster convergence speed and better optimization effect than the standard artificial bee colony algorithm. Compared with the original K-Means algorithm, the hybrid clustering algorithm proposed in this paper has better stability, fewer iterations, faster conver-gence speed and better clustering effect.