Skip to main content

Table 5 Comparison between multi-task learning and single task learning for siRNA efficacy prediction.

From: Multi-task learning for cross-platform siRNA efficacy prediction: an in-silico study

Test 3 RMSE
  Single task learning Multi-task learning
  10% 30% 50% 70% 90% 10% 30% 50% 70% 90%
E1 28.3515 24.1538 23.5544 22.8080 23.4952 27.6417 24.0150 23.5313 22.7155 23.4194
E2 28.1353 24.7949 23.0751 21.6717 20.5756 25.4531 22.0457 21.1488 20.6969 20.5423
E3 14.1021 12.7868 12.8477 12.8390 11.2925 12.4403 11.5239 11.2708 11.0255 10.0032
E4 36.7345 32.4953 30.2501 28.5389 25.5934 31.6222 27.8789 27.9831 27.5373 27.2947
E5 37.7847 31.7246 27.8395 27.2221 32.1410 37.6029 27.5771 24.0499 23.5798 24.7571
E6 37.9884 36.6409 32.8025 31.0090 27.0574 34.9948 31.9597 30.0650 28.7117 24.6019
E7 46.1408 40.6899 32.9677 34.0303 29.4516 45.3279 34.8915 30.3053 29.9185 27.0738
E8 29.4008 27.4798 26.5710 24.8380 26.7436 26.5423 24.6162 24.4261 23.7297 24.9686
E9 31.9814 15.5796 13.6068 13.8639 12.2373 35.7421 19.8070 17.2665 16.2435 13.3189
E10 56.8917 19.3907 13.4394 12.8776 11.4408 56.8917 19.1463 12.9610 12.2792 11.2242
E11 40.4318 37.2323 36.9945 34.1775 32.1200 38.9771 31.7360 31.0361 29.2156 28.6740
E12 30.7272 29.4070 33.6679 35.2603 24.8004 29.4405 24.4063 24.4616 24.8690 22.1497
E13 18.8997 18.0514 17.3333 14.1208 13.5105 18.8997 17.5524 16.4534 13.1908 10.9338
E14 34.8579 33.0815 28.7044 25.9012 25.7859 30.0917 27.8195 25.3132 24.3546 24.6832
  1. "E" denotes "Experiment". Test 3: Comparison between multi-task learning and single task learning for siRNA efficacy prediction, both trained with 10%, 30%, 50%, 70% and 90% of the data from each experiment, respectively. p-values calculated by pair t-test on multi-task learning and single task learning with different percentages of training data: 0.0268(10%); 0.0046(30%); 0.0093(50%); 0.0151(70%); 0.0389(90%).