Semi-supervised Learning for Automatic Modulation Recognition Using Haar Time–Frequency Mask and Positional–Spatial Attention

Figure 3

Overview of the HTF mask augmentation policy. The new sample pair is augmented from original unlabeled sample pair . To generate new sample (see the black line), applying a Haar mask to , inverted mask to , and summation of them. To generate new sample (see the blue line), applying another Haar mask to , inverted mask to , and summation of them.