Table 9.

Results for classification task and segmentation task with different $\lambda$ (bag size=100).

${\lambda }_{cls}/{\lambda }_{seg}$	Classification task						Segmentation task
	ACC($\text{\%}$)	Precision($\text{\%}$)	Recall($\text{\%}$)	F1 Score($\text{\%}$)	AUC($\text{\%}$)		DC($\text{\%}$)	PPV($\text{\%}$)	SEN($\text{\%}$)
1/1	97.44 $[95.72,99.50]$	96.16 $[93.46,99.58]$	97.53 $[94.30,100.0]$	96.84 $[94.24,98.95]$	97.52 $[94.92,99.19]$		95.13 ± 3.6	95.29 ± 4.5	94.97 ± 4.2
0.6/1	98.37 $[97.34,99.75]$	98.61 $[95.89,100.0]$	97.25 $[96.03,100.0]$	97.93 $[95.85,99.67]$	98.48 $[96.67,99.45]$		95.94 ± 3.3	96.06 ± 4.4	95.82 ± 3.5
0.2/1	96.86 $[95.13,98.93]$	95.95 $[93.06,99.36]$	96.37 $[92.78,98.98]$	96.16 $[93.40,98.37]$	97.03 $[93.90,98.93]$		95.42 ± 4.0	95.78 ± 4.7	95.07 ± 4.5
Learning	98.62 $[97.59,100.0]$	97.33 $[95.18,100.0]$	98.65 $[96.41,100.0]$	97.99 $[95.86,100.0]$	98.90 $[97.65,100.0]$		96.18 ± 3.7	96.26 ± 4.2	96.09 ± 3.8

Classification task results: the upper and lower bounds of 95% confidence interval are shown in [$\cdot$].

Segmentation task results: the results are shown as mean $\pm$ standard deviation.