Self-Damaging Contrastive Learning

             Ziyu Jiang 1 Tianlong Chen 2 Bobak Mortazavi 1 Zhangyang Wang 2

             Abstract                 powerful visual representations from unlabeled data. The
                                  state-of-the-art contrastive learning frameworks consistently
    The recent breakthrough achieved by contrastive
                                  benefit from using bigger models and training on more task-
    learning accelerates the pace for deploying u ...