Fig. 5From: Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural networkCS-ResCNN feature maps and representative conventional features using t-SNE. a–f Two-dimensional maps of LBP, SIFT, WT, COTE, ResCNN and CS-ResCNN methods, respectively. The red and green dots represent positive and negative samples. t-SNE, t-distributed stochastic neighbor embedding; CS-ResCNN, cost-sensitive residual convolutional neural network; WT, wavelet transformation; SIFT, scale-invariant feature transform; LBP, local binary pattern; COTE, color and texture featuresBack to article page