Table 7.

Comparison of confusion matrix value between pre-trained and trained model

Loss	Pre-train				Train
	TP	FP	TN	FN	TP	FP	TN	FN
PRAUC-loss	98,045	2,691,310	66,601,770	60,662	125,331	20,411	69,272,660	33,376
Negative-F1	15,665	84,478	69,208,610	143,043	33,447	59,797	69,233,270	125,260
Weighed-logistic	119,968	4,113,699	65,179,376	38,740	127,198	14,718	69,278,350	31,510

Columns of Pre-train is the prediction confusion matrix of different pre-trained model, and columns of Train belongs to the model after transfer learning