English(EN) The pretraining domain outweighs the training objective in setting the privacy-utility trade-off of differentially private medical image analysis

研究发现预训练域对私有医学图像AI至关重要

作者 PulseAugur 编辑部 · [1 个来源] · 2026-07-27 04:00

一篇新发表在arXiv上的研究调查了预训练数据对医学图像分析模型中隐私-效用权衡的影响。研究人员发现，与ImageNet等通用数据集相比，使用胸部X光片的预训练数据的域显著提高了差分隐私下的诊断准确性。即使预训练语料库本身是私有的，它仍然优于公共初始化，这表明在对医学成像模型应用隐私措施时，数据的来源比训练目标更关键。 AI

影响强调了在敏感医疗应用中应用差分隐私时，领域特定的预训练对于维持模型效用的重要性。

排序理由发表在arXiv上的研究论文，详细介绍了医学图像分析中隐私-效用权衡的发现。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

arXiv cs.LG TIER_1 English(EN) · Soroosh Tayebi Arasteh, Mina Farajiamiri, Mahshad Lotfinia, Behrus Hinrichs-Puladi, Jonas Bienzeisler, Mohamed Alhaskir, Mirabela Rusu, Christiane Kuhl, Sven Nebelung, Daniel Truhn · 2026-07-27 04:00

The pretraining domain outweighs the training objective in setting the privacy-utility trade-off of differentially private medical image analysis

arXiv:2601.19618v2 Announce Type: replace-cross Abstract: Differential privacy protects the patients whose images train medical imaging models, but it lowers diagnostic accuracy, and the initialization is the strongest known remedy. Practice increasingly favors large generic self…

报道来源 [1]

The pretraining domain outweighs the training objective in setting the privacy-utility trade-off of differentially private medical image analysis

相关实体

相关话题