Fig. 6
From: Deep semi-supervised learning for brain tumor classification
Performance on the MICCAI dataset, where the results were obtained from the test set averaged over 5 runs. Left: performance where GAN augmented data was added in the training set; Right: performance where no GAN augmented data was used in the training set. Red: from the proposed semi-supervised scheme; Black: from baseline-1 method; Blue: from baseline-2 method