Research Article

Clustering by Detecting Density Peaks and Assigning Points by Similarity-First Search Based on Weighted K-Nearest Neighbors Graph

Table 3

The comparison of ACC, AMI, and ARI benchmarks for 6 clustering algorithms on synthetic datasets.

AlgorithmAMIARIACCEC/ACParAMIARIACCEC/ACPar
PathbasedJain

DPC-SFSKNN0.9260.9100.9253/361.0001.0001.0002/27
DPC0.5210.4630.7423/320.6090.7130.8532/23
DBSCAN0.7810.5220.6670.05/60.8830.9850.9180.08/10
AP0.6790.4750.7833/3100.6810.8120.8822/240
FKNN-DPC0.9410.9600.9873/350.0560.1320.79310
K-means0.5680.4610.77230.4920.5770.7122

AggregationFlame

DPC-SFSKNN0.9420.9510.9637/760.8730.9340.9562/26
DPC1.0001.0001.0007/741.0001.0001.0002/25
DBSCAN0.9690.9820.9880.05/80.8670.9360.981-0.09/8
AP0.7950.7530.8417/77.70.4520.5340.8763/335
FKNN-DPC0.9950.9970.9993/381.0001.0001.0002/25
K-means0.7840.7170.78670.4180.4650.8282

DIM512DIM1024

DPC-SFSKNN1.0001.0001.00016/1681.0001.0001.00016/169
DPC1.0001.0001.00016/1621.0001.0001.00016/160.01
DBSCAN1.0001.0001.0000.3/71.0001.0001.00010/8
AP1.0001.0001.00016/16201.0001.0001.00016/1630
FKNN-DPC1.0001.0001.00016/1681.0001.0001.00016/1610
K-means0.8950.8110.85010.8680.7520.79616