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 2102.11203cs.LGcs.AIstat.ML
收藏

一种用于子种群迁移的标记传播理论

A Theory of Label Propagation for Subpopulation Shift

Princeton University(普林斯顿大学);Zhongguancun Haihua Institute for Frontier Information Technology(中关村海华前沿信息技术研究院)

Tianle Cai, Ruiqi Gao, Jason D. Lee, Qi Lei

摘要:

机器学习的核心问题之一是领域自适应。与过去的理论工作不同,我们考虑了一个新的模型,在输入或表示空间的子种群转移。在这项工作中,我们提出了一个可证明有效的框架,领域自适应标签传播的基础上。在我们的分析中,我们使用一个简单但现实的膨胀假设,提出了\citet{wei2021theoretical}。使用在源域上训练的教师分类器,我们的算法不仅传播到目标域,而且还改进了教师。通过利用现有的泛化边界,我们还获得了整个算法的端到端有限样本保证。此外,我们将我们的理论框架扩展到基于第三个未标记数据集的更一般的源到目标转移设置,这可以很容易地应用于各种学习场景。受我们理论的启发,我们将基于一致性的半监督学习方法应用于领域自适应设置,并取得了显著的改进。

英文摘要:

One of the central problems in machine learning is domain adaptation. Unlike past theoretical work, we consider a new model for subpopulation shift in the input or representation space. In this work, we propose a provably effective framework for domain adaptation based on label propagation. In our analysis, we use a simple but realistic expansion assumption, proposed in \citet{wei2021theoretical}. Using a teacher classifier trained on the source domain, our algorithm not only propagates to the target domain but also improves upon the teacher. By leveraging existing generalization bounds, we also obtain end-to-end finite-sample guarantees on the entire algorithm. In addition, we extend our theoretical framework to a more general setting of source-to-target transfer based on a third unlabeled dataset, which can be easily applied in various learning scenarios. Inspired by our theory, we adapt consistency-based semi-supervised learning methods to domain adaptation settings and gain significant improvements.

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。

↑