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

Table 2

Accuracy values of different resampling algorithms.

Algorithm	Dataset	LP-ROS	REMEDIAL-HwR-HUS	REMEDIAL-HwR-ROS	ML-DBR
LP	Yeast	0.4038	0.4179	0.4344	0.4321
LP	Enron	0.3447	0.3523	0.3605	0.3936
LP	tmc-2007	0.5929	0.6027	0.6038	0.6052
LP	cal-500	0.2069	0.1851	0.2291	0.2284
LP	Corel-16k	0.0951	0.1018	0.1196	0.1356
LP	Corel-5k	0.0864	0.1011	0.0924	0.1028
LP	Mediamill	0.4039	0.4326	0.4462	0.4697
BR	Yeast	0.4027	0.4306	0.4191	0.4286
BR	Enron	0.3134	0.3912	0.3903	0.4291
BR	tmc-2007	0.4862	0.6012	0.6068	0.6127
BR	cal-500	0.1961	0.1849	0.2336	0.2315
BR	Corel-16k	0.0958	0.1067	0.1214	0.1338
BR	Corel-5k	0.0871	0.0951	0.0894	0.0972
BR	Mediamill	0.4018	0.4276	0.4250	0.4429
ML-kNN	Yeast	0.4046	0.4219	0.4367	0.4463
ML-kNN	Enron	0.3875	0.4286	0.4322	0.4468
ML-kNN	tmc-2007	0.5328	0.6340	0.6379	0.6479
ML-kNN	cal-500	0.2494	0.2231	0.2771	0.2741
ML-kNN	Corel-16k	0.1028	0.1242	0.1266	0.1453
ML-kNN	Corel-5k	0.1020	0.1238	0.1136	0.1221
ML-kNN	Mediamill	0.4457	0.4813	0.4879	0.4908