arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2312.03318cs.LGcs.CVstat.ML
收藏

分布偏移下对比学习与自训练的互补收益

Complementary Benefits of Contrastive Learning and Self-Training Under Distribution Shift

Carnegie Mellon University(卡内基梅隆大学)

Saurabh Garg, Amrith Setlur, Zachary Chase Lipton, Sivaraman Balakrishnan, Virginia Smith, Aditi Raghunathan

AI总结:

本文系统研究自训练与对比学习结合利用无标签数据的效果,发现二者在域适应中显著互补、在半监督学习中并不协同,并通过实验和简化理论模型解释其原因。

摘要:

自训练和对比学习已成为利用无标签数据的主流技术,无论是在分布偏移(无监督域适应)情形下,还是在不存在分布偏移(半监督学习)的情形下均是如此。然而,尽管这些技术广受欢迎且彼此兼容,二者结合使用时的效果仍未得到探索。本文对这种组合进行了系统的实证研究,发现:(i)在域适应设置中,自训练和对比学习能够带来显著的互补收益;(ii)令人惊讶的是,在半监督学习设置中,这些收益并不具有协同性。在八个分布偏移数据集(例如 BREEDs、WILDS)上,我们证明组合方法比任一单独方法的准确率高出 3%–8%。随后,我们在一个简化的分布偏移模型中对这些技术进行理论分析,展示了在某些情形下,对比学习产生的特征能够为自训练提供良好的初始化,使其进一步放大收益并达到最优性能,即使任一方法单独使用都会失败。

英文摘要:

Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift (unsupervised domain adaptation) and when it is absent (semi-supervised learning). However, despite the popularity and compatibility of these techniques, their efficacy in combination remains unexplored. In this paper, we undertake a systematic empirical investigation of this combination, finding that (i) in domain adaptation settings, self-training and contrastive learning offer significant complementary gains; and (ii) in semi-supervised learning settings, surprisingly, the benefits are not synergistic. Across eight distribution shift datasets (e.g., BREEDs, WILDS), we demonstrate that the combined method obtains 3--8% higher accuracy than either approach independently. We then theoretically analyze these techniques in a simplified model of distribution shift, demonstrating scenarios under which the features produced by contrastive learning can yield a good initialization for self-training to further amplify gains and achieve optimal performance, even when either method alone would fail.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑