arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2304.00792cs.CVcs.LG
收藏

少样本微调即是无源域适应所需的一切

Few-shot Fine-tuning is All You Need for Source-free Domain Adaptation

Seoul National University of Science and Technology(首尔科学技术大学);NAVER Z Corporation(NAVER Z公司)

Suho Lee, Seungwon Seo, Jihyo Kim, Yejin Lee, Sangheum Hwang

AI总结:

论文揭示无源无监督域适应依赖无标签目标数据导致的超参数选择与分布偏移问题,提出用1-shot或3-shot标签数据微调源模型,并证明其可靠、少过拟合且在真实场景中更优。

摘要:

近年来,无源无监督域适应(SFUDA)作为一种比无监督域适应(UDA)更实际、更可行的方法而兴起,后者假设带标签的源数据始终可访问。然而,SFUDA方法的显著局限常被忽视,这限制了其在真实应用中的实用性。这些局限包括缺乏确定最优超参数的原则性方法,以及当无标签目标数据不满足闭集、与源数据标签分布相同等特定要求时出现的性能下降。所有这些局限都源于SFUDA完全依赖无标签目标数据。我们通过实验证明了现有SFUDA方法在真实场景(包括目标数据中的分布外偏移和标签分布偏移)中的局限,并验证这些方法均无法安全应用于真实环境。基于实验结果,我们主张使用少量带标签数据(例如1-shot或3-shot)微调源预训练模型,是规避SFUDA局限的实用且可靠的解决方案。与普遍看法相反,我们发现经过谨慎微调的模型即便仅用少量带标签数据训练也不会过拟合,并且因采样偏差导致的性能变化很小。我们在多个域适应基准上的实验结果表明,少样本微调方法在标准SFUDA设置下表现相当,并在真实场景中优于对比方法。代码位于https://github.com/daintlab/fewshot-SFDA。

英文摘要:

Recently, source-free unsupervised domain adaptation (SFUDA) has emerged as a more practical and feasible approach compared to unsupervised domain adaptation (UDA) which assumes that labeled source data are always accessible. However, significant limitations associated with SFUDA approaches are often overlooked, which limits their practicality in real-world applications. These limitations include a lack of principled ways to determine optimal hyperparameters and performance degradation when the unlabeled target data fail to meet certain requirements such as a closed-set and identical label distribution to the source data. All these limitations stem from the fact that SFUDA entirely relies on unlabeled target data. We empirically demonstrate the limitations of existing SFUDA methods in real-world scenarios including out-of-distribution and label distribution shifts in target data, and verify that none of these methods can be safely applied to real-world settings. Based on our experimental results, we claim that fine-tuning a source pretrained model with a few labeled data (e.g., 1- or 3-shot) is a practical and reliable solution to circumvent the limitations of SFUDA. Contrary to common belief, we find that carefully fine-tuned models do not suffer from overfitting even when trained with only a few labeled data, and also show little change in performance due to sampling bias. Our experimental results on various domain adaptation benchmarks demonstrate that the few-shot fine-tuning approach performs comparatively under the standard SFUDA settings, and outperforms comparison methods under realistic scenarios. Our code is available at https://github.com/daintlab/fewshot-SFDA .

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。

↑