arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.18031cs.CVcs.AIcs.LG
收藏

简单域泛化方法是开放域泛化的强基线

Simple Domain Generalization Methods are Strong Baselines for Open Domain Generalization

Graduate School of Environment and Information Sciences(环境与信息科学研究生院);Yokohama National University(Yokohama国立大学);Faculty of Environment(环境学系)

Masashi Noguchi, Shinichi Shirakawa

AI总结:

本文评估现有域泛化方法在开放域泛化中的表现,发现简单方法CORAL和MMD与复杂方法DAML竞争力相当,并通过集成学习和Dirichlet混合数据增强简单扩展后性能接近DAML且计算成本更低。

摘要:

在现实应用中,机器学习模型需要处理开放集识别(OSR),即在推理过程中出现未知类别,同时还要处理域偏移,即训练和推理阶段数据分布不同。域泛化(DG)旨在处理推理阶段目标域在模型训练期间不可访问的域偏移情况。开放域泛化(ODG)同时考虑DG和OSR。域增强元学习(DAML)是一种针对ODG的方法,但其学习过程复杂。相比之下,尽管已提出多种DG方法,但它们尚未在ODG场景下进行评估。在本研究中,我们全面评估了现有DG方法在ODG中的表现,并表明两种简单的DG方法——相关对齐(CORAL)和最大均值差异(MMD)——在多种情况下与DAML具有竞争力。此外,我们通过引入DAML中使用的技术(如集成学习和Dirichlet混合数据增强)提出了CORAL和MMD的简单扩展。实验评估表明,扩展后的CORAL和MMD可以以较低的计算成本达到与DAML相当的性能。这表明简单的DG方法及其简单扩展是ODG的强基线。

英文摘要:

In real-world applications, a machine learning model is required to handle an open-set recognition (OSR), where unknown classes appear during the inference, in addition to a domain shift, where the data distribution differs between the training and inference phases. Domain generalization (DG) aims to handle the domain shift situation where the target domain of the inference phase is inaccessible during the model training. Open domain generalization (ODG) considers DG and OSR. Domain-augmented meta-learning (DAML) is a method targeting ODG; however, it has a complicated learning process. By contrast, although various DG methods have been proposed, they have not been evaluated in ODG situations. In this study, we comprehensively evaluate the existing DG methods in ODG and show that the two simple DG methods, CORrelation ALignment (CORAL) and maximum mean discrepancy (MMD), are competitive with DAML in several cases. In addition, we propose simple extensions of CORAL and MMD by introducing the techniques used in DAML, such as ensemble learning and Dirichlet mixup data augmentation. The experimental evaluation demonstrates that the extended CORAL and MMD can perform comparably to DAML with lower computational costs. This suggests that the simple DG methods and their simple extensions are strong baselines for ODG.

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。

↑