A Decoupling and Bidirectional Resampling Method for Multilabel Classification of Imbalanced Data with Label Concurrence

Table 4

Micro-F values of different resampling algorithms.

Algorithm	Dataset	LP-ROS	REMEDIAL-HwR-HUS	REMEDIAL-HwR-ROS	ML-DBR
LP	Yeast	0.5737	0.6774	0.6743	0.6763
LP	Enron	0.5581	0.5649	0.6016	0.6238
LP	tmc-2007	0.6125	0.7033	0.7072	0.7103
LP	cal-500	0.4279	0.4213	0.4316	0.4424
LP	Corel-16k	0.2374	0.2867	0.3416	0.3652
LP	Corel-5k	0.2229	0.2752	0.2433	0.2809
LP	Mediamill	0.5136	0.5625	0.5593	0.5671
BR	Yeast	0.5917	0.6667	0.6753	0.6742
BR	Enron	0.5784	0.5939	0.6016	0.6429
BR	tmc-2007	0.6379	0.7221	0.7230	0.7248
BR	cal-500	0.4213	0.4652	0.4679	0.4956
BR	Corel-16k	0.2466	0.3455	0.3492	0.3601
BR	Corel-5k	0.2185	0.2673	0.2357	0.2710
BR	Mediamill	0.5278	0.5562	0.5641	0.5698
ML-kNN	Yeast	0.6939	0.6985	0.7011	0.7213
ML-kNN	Enron	0.6061	0.5732	0.6108	0.6805
ML-kNN	tmc-2007	0.7248	0.7225	0.7354	0.7826
ML-kNN	cal-500	0.5849	0.5320	0.6287	0.6735
ML-kNN	Corel-16k	0.4266	0.4212	0.4597	0.4675
ML-kNN	Corel-5k	0.3901	0.4670	0.4524	0.4709
ML-kNN	Mediamill	0.6817	0.7724	0.7782	0.7815