Table 3 Comparison of the various knowledge distillation model results.

From: Prediction of drug–target interactions through multi-task learning

   

ST → ST

ST → MT

MT → MT

Held-out

Test dataset

AUROC

Mean

0.708

0.722

0.720

Std

0.182

0.174

0.173

AUPRC

Mean

0.825

0.834

0.833

Std

0.216

0.209

0.209

Accuracy

Mean

0.749

0.753

0.752

Std

0.215

0.214

0.212

Threefold

Cross-validation

AUROC

Mean

0.900

0.899

0.899

Std

0.078

0.081

0.077

AUPRC

Mean

0.952

0.950

0.950

Std

0.080

0.082

0.082

Accuracy

Mean

0.879

0.878

0.879

Std

0.086

0.088

0.085

  1. Knowledge distillation models are represented as teacher → student. The left side of the arrow indicates the teacher model, and the right side of the arrow indicates the student model (ST: single-task learning model, MT: model of multi-task learning with similar tasks).
  2. Significant values are in bold.
Back to article page