arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.07544cs.CVcs.AI
收藏

用于无监督域泛化的解纠缠掩码自编码器

Disentangling Masked Autoencoders for Unsupervised Domain Generalization

National University of Singapore(新加坡国立大学);Zhejiang University(浙江大学);University of Science and Technology of China(中国科学技术大学)

An Zhang, Han Wang, Xiang Wang, Tat-Seng Chua

AI总结:

针对无监督域泛化研究不足的问题,提出解纠缠掩码自编码器DisMAE,通过非对称双分支架构提炼域不变语义特征,在多个基准数据集上取得了有竞争力的分布外泛化性能。

摘要:

域泛化(DG)旨在提升分布外(OOD)泛化能力,其核心是利用充足的监督信号学习对域偏移的不变性。然而,这类标注数据的稀缺催生了无监督域泛化(UDG)——这是一项更重要却更具挑战性的任务,模型以无监督方式在多个不同域上训练,最终在未见过的域上测试。UDG 正快速受到关注,但相关研究仍远未充分。为填补这一研究空白,我们提出一种专为 UDG 设计的新型学习框架,名为解纠缠掩码自编码器(Disentangled Masked Auto Encoder,DisMAE),旨在在无需类别标签的情况下,发掘能忠实反映内在特征与表层变化的解纠缠表示。该框架的核心是提炼域不变语义特征(这类特征无法被域分类器区分),同时过滤掉不稳定且冗余的域特定变化(例如配色方案和纹理模式)。值得注意的是,DisMAE 采用由语义编码器和轻量级变化编码器组成的非对称双分支架构进行协同训练,具备动态数据处理和表示层面的增强能力。在四个基准数据集(即 DomainNet、PACS、VLCS、Colored MNIST)上针对 DG 和 UDG 任务开展的大量实验表明,与当前最优的 DG 和 UDG 基线方法相比,DisMAE 能取得具有竞争力的 OOD 性能,这为利用大规模无标注数据提升泛化能力的潜在研究方向提供了思路。

英文摘要:

Domain Generalization (DG), designed to enhance out-of-distribution (OOD) generalization, is all about learning invariance against domain shifts utilizing sufficient supervision signals. Yet, the scarcity of such labeled data has led to the rise of unsupervised domain generalization (UDG) - a more important yet challenging task in that models are trained across diverse domains in an unsupervised manner and eventually tested on unseen domains. UDG is fast gaining attention but is still far from well-studied. To close the research gap, we propose a novel learning framework designed for UDG, termed the Disentangled Masked Auto Encoder (DisMAE), aiming to discover the disentangled representations that faithfully reveal the intrinsic features and superficial variations without access to the class label. At its core is the distillation of domain-invariant semantic features, which cannot be distinguished by domain classifier, while filtering out the domain-specific variations (for example, color schemes and texture patterns) that are unstable and redundant. Notably, DisMAE co-trains the asymmetric dual-branch architecture with semantic and lightweight variation encoders, offering dynamic data manipulation and representation level augmentation capabilities. Extensive experiments on four benchmark datasets (i.e., DomainNet, PACS, VLCS, Colored MNIST) with both DG and UDG tasks demonstrate that DisMAE can achieve competitive OOD performance compared with the state-of-the-art DG and UDG baselines, which shed light on potential research line in improving the generalization ability with large-scale unlabeled data.

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。

↑