Masking Improves Contrastive Self-Supervised Learning for ConvNets, and Saliency Tells You Where | IEEE Conference Publication | IEEE Xplore