Fig. 6.

Comparison precision (i.e., positive predictive value) and recall (i.e., sensitivity) between real and added synthetic dataset in different percentages of the training set. The blue and the red lines present the performance of the deep learning classification model trained by real data and the real data plus synthetic data, respectively. The horizontal axis label is the percentage of training data in the dataset. The vertical axis label is the score of precision and the recall with the range from 0 to 1