Skip to main content
. 2019 Feb 14;9:2058. doi: 10.1038/s41598-018-38343-3

Figure 5.

Figure 5

Visualisation of the learning process for a single cross validated fold where the training and validation accuracy for Inception-v3 and ResNet-50 improve after successive epochs; beginning to plateau after 100 epochs. Optimal validation accuracies of 96.7% and 97.6% were achieved at epochs 193 and 155 for Inception-v3 and ResNet-50, respectively.