Fig. 4From: Deep semi-supervised learning for brain tumor classificationTraining and validation curves (accuracy vs. epoch) on the two datasets. Left: TCGA dataset, where epochs 1-100 were from the pre-training and epochs 101-150 were from the refined training; Right: MICCAI dataset, where epochs 1-70 were from the pre-training and epochs 71-140 were from the refined trainingBack to article page