【摘要】 为解决传统密度峰聚类算法容易忽略低密度簇中心以及难以自动选择聚类中心的问题,提出罚处共享最近邻密度峰聚类算法。设计罚处系数,减少高密度簇中非中心点的共享最近邻局部密度值,降低低密度簇中心点被忽视的机率;采用迭代阈值法实现簇中心点的自动选择。在人工数据集、UCI真实数据集以及图像数据集上进行仿真实验,其结果表明,该算法能找到数据集的簇中心和簇数目,聚类精度优于相比较的其它算法,该算法是可行的、有效的。更多还原

【Abstract】 To solve the problem that the traditional density peak clustering algorithm easily ignores the low density cluster center and is difficult to automatically select the cluster center,a penalty shared nearest neighbor density peak clustering algorithm was proposed.The penalty coefficient was designed to reduce the local density value of the shared nearest neighbor of the non-center points in the high-density cluster,and reduce the probability of the center point being ignored in the low-density cluster.The iterative threshold method was used to realize the automatic selection of the cluster center point.The simulation results on artificial datasets,UCI real datasets and image data sets show that the proposed algorithm can find the cluster centers and the number of datasets,and its clustering accuracy is better than that of other algorithms.The algorithm is feasible and effective.更多还原