基于SH-ESD优化的密度峰值快速搜索聚类算法

Density peak fast search clustering algorithm based on SH-ESD optimization

通过高斯核函数的数学性质优化密度度量,根据SH-ESD异常检测算法特点,将显著水平α下检验统计量大于检验临界值的点当作簇中心,进行了E-DPC算法与DPC、DBSCAN、ADPC等算法的对比研究。

The density metric formula is optimized with mathematical properties of Gaussian kernel function.According to SH-ESD abnormal detecting characteristics,we take the tested values greater than the threshold as the candidate cluster center at a significant levelα.Comparison analysis are carried out among the E-DPC,DPC,DBSCAN and ADPC algorithms.