Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes

Abstract

In this paper, we introduce a geometric framework to analyze memorization in diffusion models through the sharpness of the log probability density. We mathematically justify a previously proposed score-difference-based memorization metric by demonstrating its effectiveness in quantifying sharpness. Additionally, we propose a novel memorization metric that captures sharpness at the initial stage of image generation in latent diffusion models, offering early insights into potential memorization. Leveraging this metric, we develop a mitigation strategy that optimizes the initial noise of the generation process using a sharpness-aware regularization term. The code is publicly available at https://github.com/Dongjae0324/sharpness_memorization_diffusion.

Cite

Text

Jeon et al. "Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes." Proceedings of the 42nd International Conference on Machine Learning, 2025.

Markdown

[Jeon et al. "Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes." Proceedings of the 42nd International Conference on Machine Learning, 2025.](https://mlanthology.org/icml/2025/jeon2025icml-understanding/)

BibTeX

@inproceedings{jeon2025icml-understanding,
  title     = {{Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes}},
  author    = {Jeon, Dongjae and Kim, Dueun and No, Albert},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  year      = {2025},
  pages     = {27091-27112},
  volume    = {267},
  url       = {https://mlanthology.org/icml/2025/jeon2025icml-understanding/}
}