scholar.google.com › citations
Dec 6, 2023 · Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift ( ...
Nov 28, 2023 · Overall, our findings demonstrate that under distribution shift features produced by contrastive learning can yield a good initialization for self-training to ...
People also ask
What are the benefits of contrastive learning?
What is the main advantage of self-supervised learning in visual representation tasks?
What is contrastive learning for binary classification?
Do generated data always help contrastive learning?
Figure 1: Self-training over Contrastive learning (STOC) improves over Contrastive Learning (CL) under distribution shift. (a) We observe that in SSL settings, ...
Dec 6, 2023 · Self-training and contrastive learning have emerged as leading techniques for incor- porating unlabeled data, both under distribution shift ...
May 30, 2024 · Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift ...
Our work examines the synergy between self-training and contrastive learning, finding complementary benefits under distribution shift setups and up to 8% ...
Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift (unsupervised ...
Dec 6, 2023 · Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift ...
Semi-Supervised U-statistics. 29 Feb 2024arXiv ; Complementary Benefits of Contrastive Learning and Self-Training Under Distribution Shift. 6 Dec 2023arXiv.
Complementary Benefits of Contrastive Learning and Self-Training Under Distribution Shift Saurabh Garg, Amrith Setlur, Zachary Chase Lipton, Sivaraman ...