Adaptive Diffusion Freezing: Privacy-preserving Diffusion Models Against Membership Inference Attacks

📅 2026-09-08
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
本文提出一种名为自适应扩散冻结(ADF)的新框架,通过控制不同数据子集在扩散时间步中的参与度来抵御成员推理攻击,同时平衡隐私、效用和效率。
📝 Abstract
Diffusion models have achieved remarkable success in generative tasks across various areas, however their training process raises significant privacy concerns, particularly under membership inference attacks (MIAs). Prior studies on privacy-preserving of diffusion models fail to balance privacy, utility, and efficiency. To address this gap, we propose a novel framework of privacy-preserving diffusion models, Adaptive Diffusion Freezing (ADF), which can defend against MIAs with better trade-off. By leveraging cross-timestep adaptive freezing training, ADF explicitly control the participation of different data subsets across diffusion timesteps via a mask matrix, which reduces the over-memorization and leads to more uniform model behaviors between member and nonmember samples. To construct a freezing mask matrix that effectively reduce membership leakage without unnecessarily harming generation quality, we introduce a pretraining-based risk-aware freezing policy to estimate MIA risk based on memorization tendency, and suppress the contribution of the subset-timestep pairs with higher risk. Evaluations on multiple datasets demonstrate that ADF provides effective defense performance as well as state-of-the-art privacy-utility-efficiency trade-off performance compared to various baselines.
Problem

Research questions and friction points this paper is trying to address.

Diffusion Models
Privacy-preserving
Membership Inference Attacks
Utility
Efficiency
Innovation

Methods, ideas, or system contributions that make the work stand out.

Adaptive Diffusion Freezing
membership inference attacks
cross-timestep adaptive freezing training
risk-aware freezing policy
🔎 Similar Papers
2023-12-07Network and Distributed System Security SymposiumCitations: 3
J
Jialu Guo
MIIT Key Laboratory of Data Intelligence and Management, Beihang University, Beijing, China
X
Xiao Han
MIIT Key Laboratory of Data Intelligence and Management, Beihang University, Beijing, China
Junjie Wu
Junjie Wu
Center for High Pressure Science & Technology Advanced Research
Physics