CAJ | 학술논문

K-均值算法是广泛使用的聚类算法,但该算法的聚类数目难以确定,且聚类结果对初始聚类中心比较敏感.本文提出一种基于微粒群优化聚类数目的K-均值算法,该算法采用聚类中心的坐标和通配符表示微粒位置,通过定义微粒更新公式中新的加减运算符,动态调整聚类中心的数目及坐标,此外,以改进的聚类有效性指标Davies-Bouldin准则作为适应度函数.5个人工和真实数据集的聚类结果验证了所提算法的优越性.
K-균치산법시엄범사용적취류산법,단해산법적취류수목난이학정,차취류결과대초시취류중심비교민감.본문제출일충기우미립군우화취류수목적K-균치산법,해산법채용취류중심적좌표화통배부표시미립위치,통과정의미립경신공식중신적가감운산부,동태조정취류중심적수목급좌표,차외,이개진적취류유효성지표Davies-Bouldin준칙작위괄응도함수.5개인공화진실수거집적취류결과험증료소제산법적우월성.
K-mean algorithm is a widely used clustering method, but it is difficult to determine the number of clusters; and the clustering result is sensitive to initial cluster centers. We present a novel K-mean algorithm for optimizing the number of clusters based on particle swarm optimization. The algorithm denotes the position of a particle with the coordinates of cluster centers and wildcards. The coordinates of cluster centers are dynamically adjusted by defining the new plus and new minus operators in the particle update formula. In addition, an improved Davies-Bouldin index is employed to evaluate the efficiency of a clustering result. Experimental results of 5 sets of artificial and real-world data validate the advantages of the proposed algorithm.