Abstract: The traditional K-means algorithm is sensitive to initial point and easy to fall into local optimum. In order to overcome these flaws, a novel clustering method based on the Modified Shuffled Frog Leaping Algorithm and K-means is presented. In this approach, a chaotic local search is introduced to improve the quality of the initial individual. Besides, mutation operating is joined to generate new individual. Simultaneously, a new searching strategy is presented to increase the optimization?ability. In addition, K-means algorithm is used according to the variation of the frog’s fitness variance. The experimental results show the proposed method improves the clustering performance, and has the advantages in the global search ability and convergence speed.

Key words: clustering, Shuffled Frog Leaping Algorithm（SFLA）, K-means, mutation, searching strategy

摘要： 针对K均值聚类算法存在的对初始值敏感且容易陷入局部最优的缺点，提出一种改进的混合蛙跳算法（SFLA）和K均值相结合的聚类算法。该算法通过混沌搜索优化初始解，变异操作生成新个体，在更新青蛙位置时，设计了一种新的搜索策略，提高了算法寻优能力；根据青蛙群体的适应度方差来确定K均值算法的操作时机，抑制早熟收敛。实验结果表明，改进的算法提高了聚类精度，在全局寻优能力和收敛速度方面具有优势。

关键词: 聚类, 混合蛙跳算法, K均值, 变异, 搜索策略