基于ISFLA的K均值聚类算法

K-means clustering algorithm based on ISFLA

针对K均值聚类算法和基于混合蛙跳(Shuffled Frog-Leaping Algorithm,SFLA)的K均值聚类算法的一些缺点,提出了基于改进混合蛙跳(Improved Shuffled Frog-Leaping AlgorithmI,SF-LA)的K均值聚类算法。该算法首先将生物学中吸引排斥机制应用在SFLA中,修改了更新策略,形成了ISFLA算法;再用该算法优化K均值聚类算法。理论分析和实验结果表明,该算法提高了收敛速度,有效地避免了SFLA早熟现象,从而改善了对高维复杂数据的搜索效率,仿真结果验证了该算法的可行性和有效性。

Because of the disadvantages of the classical K-means clustering algorithm and K-means cluster analysis based on SFLA,the paper proposes a novel K-means clustering based on an improved SFLA.The proposed algorithm integrated the attraction-repulsion mechanism in the field of biology into SFLA and modified updating strategy and became an improved SFLA.The ISFLA optimizes K-means clustering algorithm.The theory analysis and experimental results show that the proposed algorithm enhances convergence velocity and avoids premature convergence effectively,improving the efficiency of search for complex data.The result of testing shows its feasibility and validity.