Towards Efficient Contras
Towards Efficient Contrastive PAC Learning
Towards Efficient Contrastive PAC Learning
arXiv:2502.15962v1 Announce Type: cross
Abstract: We study contrastive learning under the PAC learning framework. While a series of recent works have shown statistical results for learning under contrastive loss, based either on the VC-dimension or Rademacher complexity, their algorithms are inhere…
