Table 2. Performance of our CNN model to predict miRNA:Target-site interactions.

Dataset partition methods	Sen.(%)	Spe.(%)	F1(%)	MCC(%)	Acc.(%)
train/evaluation/test	94.82	98.26	94.97	93.18	97.36
10-fold CV	93.95± 2.91	97.16± 1.25	94.76± 0.86	92.2± 1.02	97.11± 0.52

The performance on the test dataset (row two) and one-fold dataset (row three) unseen in the training processes was shown as sensitivity (column 2), specificity (column 3), F1-Score (column 4), MCC (column 5) and accuracy (column 6). For the 10-fold CV, the performance was shown as mean ± standard error (SE).