A Theoretical Framework for Preventing Class Collapse in Supervised Contrastive Learning

Open in new window