一种基于嵌入式的弱标记分类算法

一种基于嵌入式的弱标记分类算法

李亚重,杨有龙,仇海全

Label embedding for weak label classification

Yachong Li,Youlong Yang,Haiquan Qiu

表8 测试数据在排序损失上的预测结果

Table 8 Prediction results for test data on Rank Loss

Data set	$ρ$	LEWL	MLML	LRML	CPLST	BR
Emotions	0.3	0.446 $\pm$ 0.024	0.509 $\pm$ 0.042	0.441 $\pm$ 0.019	0.692 $\pm$ 0.052	0.731 $\pm$ 0.018
Emotions	0.7	0.457 $\pm$ 0.030	0.561 $\pm$ 0.071	0.453 $\pm$ 0.016	0.924 $\pm$ 0.023	0.915 $\pm$ 0.013
Yeast	0.3	0.385 $\pm$ 0.008	0.484 $\pm$ 0.009	0.709 $\pm$ 0.007	0.652 $\pm$ 0.008	0.679 $\pm$ 0.011
Yeast	0.7	0.401 $\pm$ 0.011	0.496 $\pm$ 0.013	0.722 $\pm$ 0.005	0.940 $\pm$ 0.011	0.933 $\pm$ 0.011
CAL500	0.3	0.461 $\pm$ 0.011	0.577 $\pm$ 0.006	0.518 $\pm$ 0.012	0.890 $\pm$ 0.014	0.885 $\pm$ 0.003
CAL500	0.7	0.457 $\pm$ 0.011	0.584 $\pm$ 0.018	0.528 $\pm$ 0.008	0.972 $\pm$ 0.008	0.969 $\pm$ 0.005
Medical	0.3	0.139 $\pm$ 0.037	0.476 $\pm$ 0.014	0.302 $\pm$ 0.019	0.345 $\pm$ 0.050	0.785 $\pm$ 0.027
Medical	0.7	0.168 $\pm$ 0.005	0.529 $\pm$ 0.016	0.347 $\pm$ 0.017	0.462 $\pm$ 0.040	0.977 $\pm$ 0.006
Langlog	0.3	0.468 $\pm$ 0.014	0.716 $\pm$ 0.017	0.629 $\pm$ 0.057	0.716 $\pm$ 0.053	0.689 $\pm$ 0.025
Langlog	0.7	0.487 $\pm$ 0.015	0.761 $\pm$ 0.021	0.625 $\pm$ 0.050	0.811 $\pm$ 0.031	0.887 $\pm$ 0.019
Enron	0.3	0.315 $\pm$ 0.038	0.500 $\pm$ 0.056	0.548 $\pm$ 0.008	0.792 $\pm$ 0.041	0.835 $\pm$ 0.041
Enron	0.7	0.349 $\pm$ 0.038	0.569 $\pm$ 0.055	0.771 $\pm$ 0.019	0.948 $\pm$ 0.015	0.967 $\pm$ 0.023
Corel5k	0.3	0.538 $\pm$ 0.001	0.745 $\pm$ 0.002	0.697 $\pm$ 0.002	0.825 $\pm$ 0.001	0.889 $\pm$ 0.001
Corel5k	0.7	0.549 $\pm$ 0.001	0.783 $\pm$ 0.004	0.721 $\pm$ 0.002	0.889 $\pm$ 0.004	0.903 $\pm$ 0.002