TABLE 1.

Results of our model and its variant models.

Metrics ( $mean\pm std$ )	MVIFMDA-noTR	MVIFMDA-noAR	MVIFMDA
AUC	$0.9063\pm 0.0350$	$0.9371\pm 0.0003$	$\mathbf{0.9396}\pm 0.0004$
AUPR	$0.2024\pm 0.0123$	$0.2591\pm 0.0009$	$\mathbf{0.2696}\pm 0.0009$
accuracy	$0.9931\pm 0.0002$	$0.9935\pm 0.0001$	$\mathbf{0.9936}\pm 0.0002$
precision	$0.2886\pm 0.0171$	$0.3420\pm 0.0073$	$\mathbf{0.3494}\pm 0.0118$
recall	$0.2515\pm 0.0173$	$0.2962\pm 0.0066$	$\mathbf{0.2982}\pm 0.0080$
F1	$0.2674\pm 0.0163$	$0.3162\pm 0.0019$	$\mathbf{0.3202}\pm 0.0006$
precision@5%	$0.0632\pm 0.0037$	$0.0696\pm 0.0002$	$\mathbf{0.0701}\pm 0.0002$
precision@10%	$0.0382\pm 0.0022$	$0.0414\pm 0.0001$	$\mathbf{0.0416}\pm 0.00003$
recall @5%	$0.6404\pm 0.0374$	$0.7045\pm 0.0018$	$\mathbf{0.7097}\pm 0.0018$
recall @10%	$0.7576\pm 0.0433$	$0.8214\pm 0.0010$	$\mathbf{0.8261}\pm 0.0006$

The bold values indicate the best values in rows.