一种基于改进人工蜂群的 K－means 聚类算法

摘要

As the K -means clustering method has disadvantages of sensitive to the initial clustering centers,poor global search ability,low accuracy and stability and poor robustness of algorithm,based on an improved Artificial Bee Colony algorithm,this paper proposes an algorithm to optimize the K -means clustering algorithm.It will construct a new fitness function and improve the food source location update formula to enhance the efficiency of iteration.The advantages of good global optimization ability,fast search and convergence rate are combined to improve the robustness of the algorithm.The improved algorithm is embedded in the WEKA platform,compared with the existing clustering algorithms in the WEKA,the better clustering results can be obtained.%针对 K －means 算法对初始的聚类中心选择敏感，全局搜索能力较差，聚类精度低以及稳定性不高，算法的鲁棒性较差等缺点，提出了一种基于改进的人工蜂群算法来对 K －means 聚类算法进行优化。算法构造了新的适应度函数，改进了食物源的位置更新公式来提高迭代效率。利用改进的人工蜂群算法良好的全局寻优能力，搜索速度快等优点，再加上 K －means 收敛速度快的优点，二者结合来提高算法的鲁棒性。将改进后的算法嵌入到 WEKA 这一数据挖掘平台中，充分利用了开源 WEKA 中的类和可视化功能，与 WEKA 中已有的聚类算法对比分析，可以获得更好的聚类结果。