发表机构
University of Virginia; Dexcom; University of Wisconsin-Madison（弗吉尼亚大学; 德克斯康公司; 威斯康星大学麦迪逊分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对差分隐私下表格-图像配对多模态合成的挑战，提出DP-TabImage框架，结合私有概率图模型与DP-SGD训练的表格条件扩散模型，在三类真实数据集上实现了表格、图像保真度与跨模态对齐的良好平衡。
AI 中文摘要
差分隐私（DP）合成已分别针对表格数据和图像数据进行了广泛研究，但许多真实数据集包含与多元表格记录配对的图像。在DP下合成此类数据极具挑战性，因为两种模态倾向于不同的隐私学习机制，同时必须保留它们之间的依赖关系。为应对这一挑战，我们提出DP-TabImage，一种用于隐私配对合成的模态专用框架。DP-TabImage利用私有概率图模型实例化分解式$p(x,y)=p_T(y)p_I(x\backslash y)$（用于多元表格分布），并使用DP-SGD训练的表格条件扩散模型实现条件图像分布。为在裁剪和带噪梯度下促进条件学习，我们进一步在私有表格-图像原型上预训练模型，将私有构建的属性条件图像与源自已私有表格模型的表格向量配对，且不产生额外隐私成本。在三个真实世界数据集上的实验表明，DP-TabImage在表格保真度、图像保真度和跨模态对齐之间实现了强平衡。我们的分析还揭示，视觉预热主要提升边缘图像保真度，而对齐的表格-图像预热对改善跨模态对应关系至关重要。我们的源代码可在GitHub仓库获取，链接为this https URL。
英文摘要
Differentially private (DP) synthesis has been extensively studied for tabular and image data separately, yet many real-world datasets contain images paired with multivariate tabular records. Synthesizing such data is particularly challenging under DP, as the two modalities favor different private learning mechanisms while their dependence must also be preserved. To address this challenge, we propose DP-TabImage, a modality-specialized framework for private paired synthesis. DP-TabImage instantiates the factorization $p(x,y)=p_T(y)p_I(x\;|\;y)$ using a private Probabilistic Graphical Model for the multivariate table distribution and a table-conditioned diffusion model trained with DP-SGD for the conditional image distribution. To facilitate conditional learning under clipped and noisy gradients, we further pretrain the model on private table-image prototypes, pairing privately constructed attribute-conditioned images with tabular vectors derived from the already private tabular model at no additional privacy cost. Experiments on three real-world datasets show that DP-TabImage achieves a strong balance among tabular fidelity, image fidelity, and cross-modal alignment. Our analysis further reveals that visual warm-up primarily improves marginal image fidelity, whereas aligned table-image warm-up is critical for improving cross-modal correspondence. Our source code is available in the GitHub repository, https://github.com/KaiChen9909/TabImage_Syn.
CommentsThis paper is about differentially private table-image data synthesis
