mlsanityData to 14 Oct 2026

Defending Diffusion Models Against Membership Inference Attacks via Higher-Order Langevin Dynamics

arXiv 2509.14225 ↗First posted 17 Sep 2025cs.LG stat.MLcited by 1

On the map: Diffusion models → Robustness, privacy and adversarial aspects

also in 1 more
Abstract (arXiv)

Recent advances in generative artificial intelligence applications have raised new data security concerns. This paper focuses on defending diffusion models against membership inference attacks. This type of attack occurs when the attacker can determine if a certain data point was used to train the model. Although diffusion models are intrinsically more resistant to membership inference attacks than other generative models, they are still susceptible. The defense proposed here utilizes critically-damped higher-order Langevin dynamics, which introduces several auxiliary variables and a joint diffusion process along these variables. The idea is that the presence of auxiliary variables mixes external randomness that helps to corrupt sensitive input data earlier on in the diffusion process. This concept is theoretically investigated and validated on a toy dataset and a speech dataset using the Area Under the Receiver Operating Characteristic (AUROC) curves and the FID metric.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models. Builds on 15 of its 19 references that are in the map. First posted on arXiv 17 Sep 2025.

Builds on

15 of 19 references.

Closest ancestors · 5

Same direction, newest first

  1. 2024Attacks and Defenses for Generative Diffusion Models: A Comprehensive Survey6 Aug
  2. 2023An Efficient Membership Inference Attack for the Diffusion Model by Proximal Initialization26 May
  3. 2023Membership Inference Attacks against Diffusion Models7 Feb
  4. 2023Are Diffusion Models Vulnerable to Membership Inference Attacks?2 Feb
  5. 2022Differentially Private Diffusion Models18 Oct

Other references · 10

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  4. 2016Deep Learning with Differential Privacycited by 2,728
  5. 2017Renyi Differential Privacycited by 650
  6. 2021Grad-TTS: A Diffusion Probabilistic Model for Text-to-Speechcited by 447
  7. 2021Score-Based Generative Modeling with Critically-Damped Langevin Diffusioncited by 233
  8. 2024Generative Modelling with High-Order Langevin Dynamicscited by 5
  9. 2024Critically Damped Third-Order Langevin Dynamicscited by 3
  10. 2025Critically-Damped Higher-Order Langevin Dynamics for Generative Modelingcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Reducing Diffusion Model Memorization with Higher Order Langevin Dynamicscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Noise as a Probe: Membership Inference Attacks on Diffusion Models Leveraging Initial Noise
  2. 2025Enhancing Membership Inference Attacks on Diffusion Models from a Frequency-Domain Perspective
  3. 2025Score-based Membership Inference on Diffusion Models
  4. 2023Membership Inference of Diffusion Models
  5. 2023Membership Inference Attacks on Diffusion Models via Quantile Regression
  6. 2023Towards More Realistic Membership Inference Attacks on Large Diffusion Models
  7. 2026DIME: Query-Efficient Framework for Membership Inference on Diffusion Models
  8. 2026Adaptive Diffusion Freezing: Privacy-preserving Diffusion Models Against Membership Inference Attacks