Table 8.

Comparison between multi-task learning and single task learning in a "mRNA" task level.

Test	RMSE
	T1	T2	T3	T4	T5	T6	T7	T8	T9	T10
Test 6	22.9156	29.7953	24.4563	20.2755	13.6265	25.5433	28.6792	28.6911	13.8089	47.9704
Test 7	22.0309	28.8772	34.4272	22.4800	29.5645	22.3986	23.4719	42.3385	16.1072	34.2505
Test 8	22.2569	29.4852	22.9905	19.1120	11.7851	23.5123	29.9718	28.4760	11.7036	37.8482
	T11	T12	T13	T14	T15	T16	T17	T18	T19	T20
Test 6	43.6353	13.9306	14.4649	5.6649	35.8113	33.6464	29.6981	29.4559	30.2422	21.0494
Test 7	35.4975	16.8432	13.0795	25.0440	26.3289	36.5158	29.9756	27.0347	26.0495	21.7607
Test 8	41.2163	18.2205	13.6913	5.7872	27.3318	27.5945	23.6955	26.5286	24.3853	16.2990

"T" denotes "Task". Test 6: Selected 50% of the data from each experiment to train a regression model, and tested the model on the remain 50% of the data of each experiment, respectively. Test 7: Scaled all the experimental labels into [0,1] and pooling together 50% of the data from each experiment to train a general model, and tested the model on the remain 50% of the data of each experiment, respectively. Test 8: Multi-task learning for siRNA efficacy prediction, trained with 50% of the data from each experiment, respectively. p-value calculated by pair t-test on Test 6 and Test 7 is 0.5900. p-value calculated by pair t-test on Test 6 and Test 8 is 0.0033.