A New Binary Adaptive Elitist Differential Evolution Based Automatic k-Medoids Clustering for Probability Density Functions

Table 1

The comparison of the properties of the proposed method and other existing nonhierarchical algorithms for CDFs.

Method	Define number of clusters	Solver	Address outliers/noise	Reach global solution	Computational time
MILXPM-CDF [13]	Given in advance	Classic GA	Not good	Medium	High
GA-CDF [14]	Given in advance	Modified GA	Not good	Medium	High
The method in [24]	Automatically defined	Based on data-driven learning mechanism	Not good if data is overlapping	Good	Low
The method in [25]	Given in advance	k-means	Not good	Medium	Low
The proposed method	Automatically defined	Binary aeDE	Good even for complex data	Good	Lower than [13, 14]