Skip to main content
. 2019 Jan 24;9:628. doi: 10.1038/s41598-018-36946-4

Figure 2.

Figure 2

The loss graph during training. The loss was defined as the cross entropy between predicted value and the actual one. With the iteration of training, the loss dramatically decreases and finally tends to zero. The loss graph is from the CNN-concat-filters model. Horizontal axis: iteration times. Vertical axis: loss value.