PracticalDG：面向混合域泛化的视觉语言模型扰动蒸馏方法
PracticalDG: Perturbation Distillation on Vision-Language Models for Hybrid Domain Generalization
AI总结：
针对现有域泛化方法默认类别一致、大模型训练开销高的问题，提出将视觉语言模型知识蒸馏到轻量模型的SCI-PD方法，同时构建HDG基准和$H^{2}$-CV指标，有效提升了多场景尤其是数据稀缺时的模型鲁棒性。
摘要：
域泛化（DG）旨在解决源域与目标域之间的分布偏移问题，现有DG方法默认源域和目标域数据共享完全相同的类别。但在实际场景中，目标域存在未见过的类别。为解决这一问题，开放集域泛化（OSDG）应运而生，且已有若干专门针对该任务的方法被提出。然而，大多数现有方法采用复杂的架构，相较于DG方法的性能提升十分有限。近期，视觉语言模型（VLM）通过微调范式被引入DG任务中，但使用大型视觉模型会带来极高的训练开销。因此，本文创新性地将VLM的知识迁移到轻量级视觉模型中，并从分数（Score）、类别（Class）和实例（Instance）三个维度引入扰动蒸馏（PD，即SCI-PD）来提升模型鲁棒性。此外，以往方法均基于类别一致、划分固定的基准测试，忽略了源域之间的差异。我们提出的新型基准混合域泛化（HDG）和新指标$H^{2}$-CV可构建多种数据划分以全面评估算法鲁棒性，实验表明以往方法在该基准下性能会出现急剧下降。大量实验证明，我们的方法在多个数据集上优于当前最优算法，尤其能提升数据稀缺场景下的模型鲁棒性。
英文摘要：
Domain Generalization (DG) aims to resolve distribution shifts between source and target domains, and current DG methods are default to the setting that data from source and target domains share identical categories. Nevertheless, there exists unseen classes from target domains in practical scenarios. To address this issue, Open Set Domain Generalization (OSDG) has emerged and several methods have been exclusively proposed. However, most existing methods adopt complex architectures with slight improvement compared with DG methods. Recently, vision-language models (VLMs) have been introduced in DG following the fine-tuning paradigm, but consume huge training overhead with large vision models. Therefore, in this paper, we innovate to transfer knowledge from VLMs to lightweight vision models and improve the robustness by introducing Perturbation Distillation (PD) from three perspectives, including Score, Class and Instance (SCI), named SCI-PD. Moreover, previous methods are oriented by the benchmarks with identical and fixed splits, ignoring the divergence between source domains. These methods are revealed to suffer from sharp performance decay with our proposed new benchmark Hybrid Domain Generalization (HDG) and a novel metric $H^{2}$-CV, which construct various splits to comprehensively assess the robustness of algorithms. Extensive experiments demonstrate that our method outperforms state-of-the-art algorithms on multiple datasets, especially improving the robustness when confronting data scarcity.
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。
