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7 测试数据在Macro F1上的预测结果
|
Table 7 Prediction results for test data on Macro F1
|
|
Data set | | LEWL | MLML | LRML | CPLST | BR |
---|
Emotions | 0.3 | 0.6060.057 | 0.5760.051 | 0.6160.036 | 0.4290.069 | 0.3890.021 | 0.7 | 0.5960.038 | 0.5230.061 | 0.6100.036 | 0.1520.033 | 0.1600.041 | Yeast | 0.3 | 0.4610.004 | 0.3340.011 | 0.3670.008 | 0.2510.009 | 0.2340.009 | 0.7 | 0.4520.006 | 0.3130.008 | 0.3660.009 | 0.0580.009 | 0.0620.004 | CAL500 | 0.3 | 0.2340.012 | 0.0730.007 | 0.2120.011 | 0.0350.005 | 0.0460.003 | 0.7 | 0.2320.008 | 0.0660.008 | 0.2050.010 | 0.0080.002 | 0.0150.002 | Medical | 0.3 | 0.3580.032 | 0.3110.012 | 0.3250.002 | 0.3100.023 | 0.0980.018 | 0.7 | 0.3060.020 | 0.2580.008 | 0.2870.019 | 0.1710.029 | 0.010.003 | Langlog | 0.3 | 0.4480.018 | 0.1640.011 | 0.3890.028 | 0.1710.025 | 0.1850.014 | 0.7 | 0.4340.017 | 0.2320.009 | 0.3740.029 | 0.0480.009 | 0.0590.0120. | Enron | 0.3 | 0.1660.011 | 0.0650.013 | 0.1200.015 | 0.0630.009 | 0.0420.008 | 0.7 | 0.1490.010 | 0.0510.014 | 0.1010.016 | 0.0220.004 | 0.0130.011 | Corel5k | 0.3 | 0.0250.001 | 0.0180.005 | 0.0200.003 | 0.0100.003 | 0.0090.003 | 0.7 | 0.0190.001 | 0.0150.006 | 0.0160.003 | 0.0070.004 | 0.0060.002 |
|
|
|