Table 3 Performance comparison of using different T-loss in JLAN
From: JLAN: medical code prediction via joint learning attention networks and denoising mechanism
config | MIMIC-III-full | MIMIC-III-50 | ||
|---|---|---|---|---|
Micro-F1 | Macro-F1 | Micro-F1 | Macro-F1 | |
T-loss=0.05 | 0.542 | 0.061 | 0.623 | 0.571 |
T-loss=0.1 | 0.557 | 0.068 | 0.626 | 0.574 |
T-loss=0.15 | 0.556 | 0.068 | 0.627 | 0.573 |
T-loss=0.2 | 0.547 | 0.064 | 0.625 | 0.573 |