Skip to main content
. Author manuscript; available in PMC: 2020 Jan 9.
Published in final edited form as: IEEE Access. 2019 Jan 9;7:11093–11104. doi: 10.1109/ACCESS.2019.2891970

Fig. 5.

Fig. 5.

MNIST results for 100 and 1000 labeled samples for training, compared with Semi-supervised embedding [15], AtlasRBF [43], the Ladder network [9], Improved GAN [44], Virtual adversarial training (VAT) [45], Mutual exclusivity [10] and Learning by association [16]. The best results in each column are shown in bold text.