Fig. 4From: A semi-supervised deep learning approach for predicting the functional effects of genomic non-coding variationsThe performance of the proposed model trained with a certain cell-line dataset and evaluated with the validation datasets of other cell lines by AUC values (a) and Accuracy values (b). The bars represent the standard errors in fivefold cross validation and p-values were calculated by two-tailed t-test; AUC, area under the ROC (receiver operating characteristic) curve; SSL_dnn, semi-supervised learning by a deep neural network with pseudo labelsBack to article page