大数据挖掘中的混合差分进化K-Means无监督聚类算法

Hybrid Differential Evolution K-Means Unsupervised Clustering Algorithm in Big Data Mining

K-Means无监督聚类算法是现有聚类算法中最为典型的划分算法。针对K-Means聚类算法初始参数依赖性较高且聚类结果稳定性较差的问题,提出了一种改进的混合差分进化算法,并将混合差分进化算法引入K-Means聚类中。通过个体适值函数把种群视为2个子种群的混合体,并按照不同的变异策略和参数对2个子种群分别进行动态更新,提高了获取全局最优的概率。实验结果表明:相比K-Means聚类算法、基于差分进化的K-均值聚类算法,所提出方法能够有效提高聚类质量和收敛速度,较好地解决了K-Means聚类算法容易陷入局部最优陷阱的问题。

K-Means unsupervised clustering algorithm is the most typical partitioning algorithm in existing clustering algorithms.Aiming at the problem that K-Means clustering algorithm has high initial parameter dependence and poor stability of clustering results,an improved hybrid differential evolution algorithm is proposed and the hybrid differential evolution algorithm is introduced into K-Means clustering.The population is regarded as a mixture of two sub-populations by the individual fitness function,and the two sub-populations are dynamically updated according to different mutation strategies and parameters,which improves the probability of obtaining global optimality.The proposed algorithm solves the problem that the K-Means clustering algorithm is easy to fall into the local optimal trap.The experiments show that compared with K-Means clustering algorithm and K-Means clustering algorithm based on differential evolution,the proposed method can effectively improve the clustering quality and convergence speed.