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 2301.13428cs.CVcs.LG
收藏

对比与聚类:面向无源域适应的邻域对表示学习

Contrast and Clustering: Learning Neighborhood Pair Representation for Source-free Domain Adaptation

Jiaxing University(嘉兴学院);Zhejiang Normal University(浙江师范大学);Zhejiang Sci-Tech University(浙江理工大学)

Yuqi Chen, Xiangbin Zhu, Yonggang Li, Yingjian Li, Haojie Fang

AI总结:

针对无源域数据的无监督域适应场景,本文从对比学习视角出发,通过最近邻聚类、扩展邻域构造难负样本对、结合噪声对比估计的方法学习域不变特征,在三个基准数据集上取得了优于现有最优方法的效果。

摘要:

无监督域适应利用来自不同分布的源域数据,解决无标签目标域的数据分类问题。然而,传统方法需要访问源数据,这往往会引发数据隐私方面的担忧。本文研究了一种更具实用性但也更具挑战性的场景:源域数据不可获取,且目标域数据无标签。具体而言,我们从对比学习的视角出发解决域差异问题。本研究的核心思路是通过三个步骤学习域不变特征:1)在原始特征空间中直接基于最近邻执行聚类;2)在不引入额外计算复杂度的前提下,通过扩展邻域构造真正的难负样本对;3)结合噪声对比估计理论获得计算优势。我们在三个常用基准数据集VisDA、Office-Home和Office-31上开展了细致的消融研究与大量实验,结果表明,与其他现有最优方法相比,本文所提方法性能更优。

英文摘要:

Unsupervised domain adaptation uses source data from different distributions to solve the problem of classifying data from unlabeled target domains. However, conventional methods require access to source data, which often raise concerns about data privacy. In this paper, we consider a more practical but challenging setting where the source domain data is unavailable and the target domain data is unlabeled. Specifically, we address the domain discrepancy problem from the perspective of contrastive learning. The key idea of our work is to learn a domain-invariant feature by 1) performing clustering directly in the original feature space with nearest neighbors; 2) constructing truly hard negative pairs by extended neighbors without introducing additional computational complexity; and 3) combining noise-contrastive estimation theory to gain computational advantage. We conduct careful ablation studies and extensive experiments on three common benchmarks: VisDA, Office-Home, and Office-31. The results demonstrate the superiority of our methods compared with other state-of-the-art works.

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。

↑